Cable Sal
Member
- Messages
- 70
- Likes
- 7
Cable Sal
Member
- Messages
- 70
- Likes
- 7
Hi y'all,
It's occurred to me that I'm building my model based on some key assumptions about spreadbetting, and I'd really appreciate a multitude of other eyes looking over these assumptions. If I've got anything wrong here, it'll throw everything else out.
Price conversions
I use Yahoo to give me a market bid and market ask price for a stock, then I apply a SB spread increase. IG Index quote 0.1% either side for share betting, so I apply that to my market prices. Question 1: Is Yahoo Finance a decent proxy for the "market price" that IG Index say they apply their 0.1% either side spread to?
Example: Yahoo quotes bid = 299.70, ask = 300.80 for Stock H.
SB equivalent price: bid = Yahoo bid - (Yahoo bid x 0.1%), ask = Yahoo ask + (Yahoo ask x 0.1%)
SB equivalent price: bid = 299.70 - (299.70 x 0.1%), ask = 300.80 + (300.80 x 0.1%)
SB equivalent price: bid = 299.40, ask = 301.10
Question 2: Have I applied the 0.1% spread increase in such a way that it reasonably reflects SB pricing?
To limit the number of variables that I'm working with initially, I'm basing everything on guaranteed risk, which means that (according to IG Index), I have a typical premium of 0.7% included in my opening price, and a minimum stop gap of 7.5%. I'm using IG Index medium volatility for consistency, at some point I'll compare other volatilities and SB companies. To apply the 0.7% premium I compound it with the 0.1% SB spread.
Example: Yahoo quotes bid = 299.70, ask = 300.80 for Stock H.
If I'm going long, I add the premium to the ask
SB equivalent opening ask = (Yahoo ask + (Yahoo ask x 0.1%)) + ((Yahoo ask + (Yahoo ask x 0.1%)) x 0.7%)
SB equivalent opening ask = (300.80 + (300.80 x 0.1%)) + ((300.80 + (300.80 x 0.1%)) x 0.7%)
SB equivalent opening ask = 303.21
If I'm going short, I subtract the premium from the bid
SB equivalent opening bid = (Yahoo bid - (Yahoo bid x 0.1%)) - ((Yahoo bid - (Yahoo bid x 0.1%)) x 0.7%)
SB equivalent opening bid = (299.70 - (299.70 x 0.1%)) - ((299.70 - (299.70 x 0.1%)) x 0.7%)
SB equivalent opening bid = 297.30
Question 3: Am I right to compound the 0.1% and 0.7%? I'm assuming that the premium would be applied to the SB's own price, not the market price.
Minimum stop on a guaranteed risk
Next, to calculate the minimum stop gap of 7.5%, I first calculate what price that would be on the SB platform, and then convert it back to a Yahoo equivalent so I can use it in my experimental tracking.
Example: Yahoo quotes bid = 299.70, ask = 300.80 for Stock H.
SB stop price long = SB bid price - (SB bid price x 7.5%)
SB stop price long = 299.40 - (299.40 x 7.5%)
SB stop price long = 276.95
Yahoo price which would stop out that trade = SB stop price long + (SB stop price long x 0.1%)
Yahoo price which would stop out that trade = 276.95 + (276.95 x 0.1%)
Yahoo price which would stop out that trade = 277.22
SB stop price short = SB ask price + (SB ask price x 7.5%)
SB stop price short = 301.10 + (301.10 x 7.5%)
SB stop price short = 323.68
Yahoo price which would stop out that trade = SB stop price short - (SB stop price short x 0.1%)
Yahoo price which would stop out that trade = 323.68 - (323.68 x 0.1%)
Yahoo price which would stop out that trade = 323.36
Question 4: Am I measuring the so called stop gap from the right baseline? I'm assuming 7.5% stop gap is 7.5% below the SB's own bid price if I'm going long and 7.5% above the SB's own ask price if I'm going short. This may be particular to IG Index.
Risk and reward
I calculate how much risk I'm exposed to by working out how much my account would be down if I stopped out for the full amount using the formulae above. I work out the change in price, which is the difference between what I opened at and what I closed at, and then multiply that by my bet size.
Example: SB quotes bid = 299.40, ask = 301.10 for Stock H, I bet £10 and get stopped out
Going long, change in price = SB opening ask including premium - SB stop price long
Going long, change in price = 303.21 - 276.95
Going long, change in price = 26.26
Loss on long stop = change in price x bet size
Loss on long stop = 26.26 x 10
Loss on long stop = £262.63
Going short, change in price = SB stop price short - SB opening bid including premium
Going short, change in price = 323.68 - 297.30
Going short, change in price = 26.38
Loss on short stop = change in price x bet size
Loss on short stop = 26.38 x 10
Loss on short stop = £263.79
Question 5: Leaving aside for a moment better ways of calculating where to put my stop, am I right in assuming that my risk exposure is equal to my loss on a full bet stop out? It's a guaranteed stop, so there's no slippage to consider in this example.
And finally (many thanks if you've stuck through the sums this far), I'm looking to get a risk:reward ratio of 1:2. For the purposes of simplicity I'll ignore scale out here. To achieve 1:2 risk:reward I need to make twice the value when I win that I lose when I stop out. To calculate that I double my stopped out loss value and divide that through by the bet size to give me the change in price needed to generate the profit. I take the change in price needed and add/subtract it from the opening price including premium to find the closing price that the SB would need to quote for me to take a profit of twice my risk. I then convert this back to a Yahoo price so that I can use it in my experimental tracking.
Example: Yahoo quotes bid = 299.70, ask = 300.80 for Stock H, my bet size = £10
Therefore: SB equivalent price: bid = 299.40, ask = 301.10
And: SB equivalent opening ask = 303.21
And: SB equivalent opening bid = 297.30
And: Loss on long stop = £262.63
And: Loss on short stop = £263.79
Profit value needed going long = 2 x Loss on long stop
Profit value needed going long = 2 x £262.63
Profit value needed going long = £525.26
Long change in price needed = profit value needed / bet size
Long change in price needed = £525.26 / £10
Long change in price needed = 52.53
New SB long price needed = SB equivalent opening ask + long change in price
New SB long price needed = 303.21 + 52.53
New SB long price needed = 355.73
Yahoo equivalent price needed = new SB long price - (new SB long price x 0.1%)
Yahoo equivalent price needed = 355.73 - (new SB long price x 0.1%)
Yahoo equivalent price needed = 355.38
Profit value needed going short = 2 x Loss on short stop
Profit value needed going short = 2 x £263.79
Profit value needed going short = £527.58
Short change in price needed = profit value needed / bet size
Short change in price needed = £527.58 / £10
Short change in price needed = 52.76
New SB short price needed = SB equivalent opening bid - short change in price
New SB short price needed = 297.30 + 52.76
New SB short price needed = 244.55
Yahoo equivalent price needed = new SB short price + (new SB short price x 0.1%)
Yahoo equivalent price needed = 244.55 + (new SB short price x 0.1%)
Yahoo equivalent price needed = 244.79
Question 6: Have I got the concept of a risk:reward ratio right, doubling the value of my stop losses to achieve 1:2?
Gold stars to anybody who managed to read through that lot, it looks like a school exam paper. I'd much appreciate the help of anybody prepared to have a pop at the questions, but fully understand if you lost the will to live half way through 😱nline2lo
Cheers,
Sal
(odds on I have to come back and edit a typo in that lot, there's no way I could have got all those numbers right first time!)
1st edit: just a formatting thing
It's occurred to me that I'm building my model based on some key assumptions about spreadbetting, and I'd really appreciate a multitude of other eyes looking over these assumptions. If I've got anything wrong here, it'll throw everything else out.
Price conversions
I use Yahoo to give me a market bid and market ask price for a stock, then I apply a SB spread increase. IG Index quote 0.1% either side for share betting, so I apply that to my market prices. Question 1: Is Yahoo Finance a decent proxy for the "market price" that IG Index say they apply their 0.1% either side spread to?
Example: Yahoo quotes bid = 299.70, ask = 300.80 for Stock H.
SB equivalent price: bid = Yahoo bid - (Yahoo bid x 0.1%), ask = Yahoo ask + (Yahoo ask x 0.1%)
SB equivalent price: bid = 299.70 - (299.70 x 0.1%), ask = 300.80 + (300.80 x 0.1%)
SB equivalent price: bid = 299.40, ask = 301.10
Question 2: Have I applied the 0.1% spread increase in such a way that it reasonably reflects SB pricing?
To limit the number of variables that I'm working with initially, I'm basing everything on guaranteed risk, which means that (according to IG Index), I have a typical premium of 0.7% included in my opening price, and a minimum stop gap of 7.5%. I'm using IG Index medium volatility for consistency, at some point I'll compare other volatilities and SB companies. To apply the 0.7% premium I compound it with the 0.1% SB spread.
Example: Yahoo quotes bid = 299.70, ask = 300.80 for Stock H.
If I'm going long, I add the premium to the ask
SB equivalent opening ask = (Yahoo ask + (Yahoo ask x 0.1%)) + ((Yahoo ask + (Yahoo ask x 0.1%)) x 0.7%)
SB equivalent opening ask = (300.80 + (300.80 x 0.1%)) + ((300.80 + (300.80 x 0.1%)) x 0.7%)
SB equivalent opening ask = 303.21
If I'm going short, I subtract the premium from the bid
SB equivalent opening bid = (Yahoo bid - (Yahoo bid x 0.1%)) - ((Yahoo bid - (Yahoo bid x 0.1%)) x 0.7%)
SB equivalent opening bid = (299.70 - (299.70 x 0.1%)) - ((299.70 - (299.70 x 0.1%)) x 0.7%)
SB equivalent opening bid = 297.30
Question 3: Am I right to compound the 0.1% and 0.7%? I'm assuming that the premium would be applied to the SB's own price, not the market price.
Minimum stop on a guaranteed risk
Next, to calculate the minimum stop gap of 7.5%, I first calculate what price that would be on the SB platform, and then convert it back to a Yahoo equivalent so I can use it in my experimental tracking.
Example: Yahoo quotes bid = 299.70, ask = 300.80 for Stock H.
SB stop price long = SB bid price - (SB bid price x 7.5%)
SB stop price long = 299.40 - (299.40 x 7.5%)
SB stop price long = 276.95
Yahoo price which would stop out that trade = SB stop price long + (SB stop price long x 0.1%)
Yahoo price which would stop out that trade = 276.95 + (276.95 x 0.1%)
Yahoo price which would stop out that trade = 277.22
SB stop price short = SB ask price + (SB ask price x 7.5%)
SB stop price short = 301.10 + (301.10 x 7.5%)
SB stop price short = 323.68
Yahoo price which would stop out that trade = SB stop price short - (SB stop price short x 0.1%)
Yahoo price which would stop out that trade = 323.68 - (323.68 x 0.1%)
Yahoo price which would stop out that trade = 323.36
Question 4: Am I measuring the so called stop gap from the right baseline? I'm assuming 7.5% stop gap is 7.5% below the SB's own bid price if I'm going long and 7.5% above the SB's own ask price if I'm going short. This may be particular to IG Index.
Risk and reward
I calculate how much risk I'm exposed to by working out how much my account would be down if I stopped out for the full amount using the formulae above. I work out the change in price, which is the difference between what I opened at and what I closed at, and then multiply that by my bet size.
Example: SB quotes bid = 299.40, ask = 301.10 for Stock H, I bet £10 and get stopped out
Going long, change in price = SB opening ask including premium - SB stop price long
Going long, change in price = 303.21 - 276.95
Going long, change in price = 26.26
Loss on long stop = change in price x bet size
Loss on long stop = 26.26 x 10
Loss on long stop = £262.63
Going short, change in price = SB stop price short - SB opening bid including premium
Going short, change in price = 323.68 - 297.30
Going short, change in price = 26.38
Loss on short stop = change in price x bet size
Loss on short stop = 26.38 x 10
Loss on short stop = £263.79
Question 5: Leaving aside for a moment better ways of calculating where to put my stop, am I right in assuming that my risk exposure is equal to my loss on a full bet stop out? It's a guaranteed stop, so there's no slippage to consider in this example.
And finally (many thanks if you've stuck through the sums this far), I'm looking to get a risk:reward ratio of 1:2. For the purposes of simplicity I'll ignore scale out here. To achieve 1:2 risk:reward I need to make twice the value when I win that I lose when I stop out. To calculate that I double my stopped out loss value and divide that through by the bet size to give me the change in price needed to generate the profit. I take the change in price needed and add/subtract it from the opening price including premium to find the closing price that the SB would need to quote for me to take a profit of twice my risk. I then convert this back to a Yahoo price so that I can use it in my experimental tracking.
Example: Yahoo quotes bid = 299.70, ask = 300.80 for Stock H, my bet size = £10
Therefore: SB equivalent price: bid = 299.40, ask = 301.10
And: SB equivalent opening ask = 303.21
And: SB equivalent opening bid = 297.30
And: Loss on long stop = £262.63
And: Loss on short stop = £263.79
Profit value needed going long = 2 x Loss on long stop
Profit value needed going long = 2 x £262.63
Profit value needed going long = £525.26
Long change in price needed = profit value needed / bet size
Long change in price needed = £525.26 / £10
Long change in price needed = 52.53
New SB long price needed = SB equivalent opening ask + long change in price
New SB long price needed = 303.21 + 52.53
New SB long price needed = 355.73
Yahoo equivalent price needed = new SB long price - (new SB long price x 0.1%)
Yahoo equivalent price needed = 355.73 - (new SB long price x 0.1%)
Yahoo equivalent price needed = 355.38
Profit value needed going short = 2 x Loss on short stop
Profit value needed going short = 2 x £263.79
Profit value needed going short = £527.58
Short change in price needed = profit value needed / bet size
Short change in price needed = £527.58 / £10
Short change in price needed = 52.76
New SB short price needed = SB equivalent opening bid - short change in price
New SB short price needed = 297.30 + 52.76
New SB short price needed = 244.55
Yahoo equivalent price needed = new SB short price + (new SB short price x 0.1%)
Yahoo equivalent price needed = 244.55 + (new SB short price x 0.1%)
Yahoo equivalent price needed = 244.79
Question 6: Have I got the concept of a risk:reward ratio right, doubling the value of my stop losses to achieve 1:2?
Gold stars to anybody who managed to read through that lot, it looks like a school exam paper. I'd much appreciate the help of anybody prepared to have a pop at the questions, but fully understand if you lost the will to live half way through 😱nline2lo
Cheers,
Sal
(odds on I have to come back and edit a typo in that lot, there's no way I could have got all those numbers right first time!)
1st edit: just a formatting thing
barjon
Legendary member
- Messages
- 10,752
- Likes
- 1,863
Sal
Whilst I understand your desire to model all this I do think you're over complicating things. A couple of points.
1. Don't forget that the spread is not static, but variable. Even on liquid equities it will widen significantly as the market tries to find a price - like at the opening.
2. If you're going to spreadbet why bother trying to relate the market price to the spreadbet price. Why not just bother about IG's price. If you confine yourself to properly liquid equities the price will be close enough.
3. Similarly, if you confine yourself to good size, liquid equities - say most of FTSE100 and a few others - then why throw money away on a guaranteed stop. It's like paying an insurance premium to protect against flood when you live half way up a hill. Possible but highly unlikely.
jon
Whilst I understand your desire to model all this I do think you're over complicating things. A couple of points.
1. Don't forget that the spread is not static, but variable. Even on liquid equities it will widen significantly as the market tries to find a price - like at the opening.
2. If you're going to spreadbet why bother trying to relate the market price to the spreadbet price. Why not just bother about IG's price. If you confine yourself to properly liquid equities the price will be close enough.
3. Similarly, if you confine yourself to good size, liquid equities - say most of FTSE100 and a few others - then why throw money away on a guaranteed stop. It's like paying an insurance premium to protect against flood when you live half way up a hill. Possible but highly unlikely.
jon
B
Black Swan
0007
Senior member
- Messages
- 2,377
- Likes
- 664
Sal,
I agree with Barjon's comments. As an academic exercise what you're doing is very interesting but for the purposes of arriving at a profitable system I'm not convinced its necessary at this stage of your investigations.
One of the problems is always data: Yahoo's is not "cleaned up" but it's free. You can get supposedly accurate data from other providers at a price. But all data is a sort of average - depends on the market maker & this is what SB Co's do. Spreads vary according to time of day (and presumably the balance of the MM's 4 humors). So in the great scheme of things I wonder if it is that important. In addition to that there is slippage (which may not always be present - yet another variable) so any attempt at an "accurate" price is not a trivial matter.
I like the financial statistician Millard's method where he takes an artificial set of data (of which the outcome is known) and applies his system to it. If that proves a workable system (ie one that reflects the known results) it is then applied to random data. Just a form of backtesting I suppose. I wonder if you could arrange your data like this so that you just set fixed but known realistic parameters on the data - you at least know what you are dealing with; I suspect this is what you are trying to do anyway.
While I agree with Barjon's comments on guaranteed stops in the real world for liquid instruments, I think in your testing case it's another way of taking a variable out of the equation and will help to see if your system works well. On the other hand it's difficult for your system to model the human factor which, unless your trading is to be 100% mechanical automatic, will most likely have a great effect on your trading (htink you alluded to that in your 1st post).
I think you have to lookat this pragmatically - there are very clever people who apply maths to the markets in ways I could never understand and likely produce good results (they don't pay Quants loads of dosh just to sit and drink coffee all day do they?) - I think a good place to start is on the back of an envelope (or fag packet/beermat/etc) to get the ideas right with very simple calculations which you can "improve" (ie compliacte) later.
I think I would just make "reasonable" assumptions on your data params and take it from there.
I agree with Barjon's comments. As an academic exercise what you're doing is very interesting but for the purposes of arriving at a profitable system I'm not convinced its necessary at this stage of your investigations.
One of the problems is always data: Yahoo's is not "cleaned up" but it's free. You can get supposedly accurate data from other providers at a price. But all data is a sort of average - depends on the market maker & this is what SB Co's do. Spreads vary according to time of day (and presumably the balance of the MM's 4 humors). So in the great scheme of things I wonder if it is that important. In addition to that there is slippage (which may not always be present - yet another variable) so any attempt at an "accurate" price is not a trivial matter.
I like the financial statistician Millard's method where he takes an artificial set of data (of which the outcome is known) and applies his system to it. If that proves a workable system (ie one that reflects the known results) it is then applied to random data. Just a form of backtesting I suppose. I wonder if you could arrange your data like this so that you just set fixed but known realistic parameters on the data - you at least know what you are dealing with; I suspect this is what you are trying to do anyway.
While I agree with Barjon's comments on guaranteed stops in the real world for liquid instruments, I think in your testing case it's another way of taking a variable out of the equation and will help to see if your system works well. On the other hand it's difficult for your system to model the human factor which, unless your trading is to be 100% mechanical automatic, will most likely have a great effect on your trading (htink you alluded to that in your 1st post).
I think you have to lookat this pragmatically - there are very clever people who apply maths to the markets in ways I could never understand and likely produce good results (they don't pay Quants loads of dosh just to sit and drink coffee all day do they?) - I think a good place to start is on the back of an envelope (or fag packet/beermat/etc) to get the ideas right with very simple calculations which you can "improve" (ie compliacte) later.
I think I would just make "reasonable" assumptions on your data params and take it from there.
Cable Sal
Member
- Messages
- 70
- Likes
- 7
I'm not sure if this is a private joke or something I should be concerned about, whichever, I don't get it. Sorry Black Swan, I'll have to ignore it until I know enough to laugh with you.
Jon and 0007, thanks for your input, you've both given me food for thought, and a bit of reassurance that neither of you have said anything like "Sal, you donkey, you've subtracted the result from the bid instead of adding it to the ask", or some such thing.
Here's my schedule, partly timed around the peaks and troughs of my day job, partly around some big events in my private life, and partly an understanding how my head works and how to get the best out of it.
Aug10 - Oct10 Pull it apart and see how it works (play with spreadsheets and data)
Oct10 - Dec10 Decide on my rules of engagement and flesh out a trading plan
Dec10 - Jan10 Faux-paper trading (strict testing of the plan, but with freely available data)
Jan11 - Feb11 Open account(s) with SB and paper trade on their platform, adjusting trading plan for reality
Feb11 - Apr11 Sometime around this point, I'll put real money behind my trading plan.
So, apologies for those who must find it frustrating that, while I'm not trying to reinvent the wheel, I'm still insistant on pulling the spokes off to see how they hold the rim on. I'll have got through this phase by October.
I'm making it slightly more difficult for myself by holding off opening a SB account until next year, which means I don't have access to SB prices, which is why I model them instead. I know if I open an account now it'll distract me from developing a good trading plan. I can invest time to learn stuff, or I can invest money to learn stuff. No doubt, even with all this preparation I'll still blow a good amount of my initial capital when I get going, but time is definitely a cheaper investment than money while I'm a noob.
I agree with you both about guaranteed stops - I don't believe they are value for money as insurance, as you say Jon - and I'm only using it to limit the variables, as you suggested 0007. The guaranteed stops are high on the list for removal and replacement with a slippage variable and stop placement methodology.
Millard's method? I'm interested. I don't have a background in economics, I guess that's why I'm using methodologies designed to make planes fly in uncertain air densities, lol, I know how to do that. I've just had a search on the forum and found who you're talking about, so I shall investigate more.
By the way Jon, I love your Sunday debates, very educational. 👍
Cheers,
Sal
Cable Sal
Member
- Messages
- 70
- Likes
- 7
Sorry, let me rephrase that - I'm trying to take as much belief / disbelief / opinion / conjecture as possible out of my experiment at the moment. My early results are indicating that guaranteed stops are poor value for money. I'm planning on comparing it with the same model but using slippage variables and some stop placement methodologies. First step is to come up a reasonable slippage variable, there's loads of stop placement methodologies that I'd like to try out, including one mentioned earlier in this thread. To assess the value of the risk limitation that comes with guaranteed stops, I'll use standard QRA methodology.
Cable Sal
Member
- Messages
- 70
- Likes
- 7
Okay, things are looking good - I'm able to track the profit / loss of a simulated spreadbet account based on my selection of 11 FTSE 250, ranging in price from 12p to 730p, some going up, some going down and one or two that are fairly range bound. The spread, both of the underlying stock and the additional spread of the SB is accounted for.
Time to experiment
It's still based on a guaranteed stop, and that's the first thing that I want to replace by introducing extra variables. Can anybody help me here? What's typical for SB slippage? I'm guessing there's no definitive answer - how long is a piece of string - but there'll be a standard deviation of slippage. Just knowing the scale, the order, that I'm working with would be a good help.
If I'm simulating placing a non-guaranteed stop, and that stop is hit, do I allow for 30% slippage - or is that a bit excessive compared to what I might normally experience. Is 10% or 5% more typical?
I'm assuming that I would weight the slippage by the price - but is volume a factor? Should I be looking to apply a higher slippage percentage when volume of trade is low and price change is steep? Is the spread a factor? A wider spread being indicative of a greater liklihood of slippage? This is all easy stuff to build in, but tricky to guess what parameters I should work with.
As always, suggestions are most appreciated. Maybe this is a question I need to raise in the SB subforum?
Cheers,
Sal
Time to experiment
It's still based on a guaranteed stop, and that's the first thing that I want to replace by introducing extra variables. Can anybody help me here? What's typical for SB slippage? I'm guessing there's no definitive answer - how long is a piece of string - but there'll be a standard deviation of slippage. Just knowing the scale, the order, that I'm working with would be a good help.
If I'm simulating placing a non-guaranteed stop, and that stop is hit, do I allow for 30% slippage - or is that a bit excessive compared to what I might normally experience. Is 10% or 5% more typical?
I'm assuming that I would weight the slippage by the price - but is volume a factor? Should I be looking to apply a higher slippage percentage when volume of trade is low and price change is steep? Is the spread a factor? A wider spread being indicative of a greater liklihood of slippage? This is all easy stuff to build in, but tricky to guess what parameters I should work with.
As always, suggestions are most appreciated. Maybe this is a question I need to raise in the SB subforum?
Cheers,
Sal
0007
Senior member
- Messages
- 2,377
- Likes
- 664
Sal,
I can only give you my limited experience on this using IG Index - limited because it may not be typical of other users but more importantly because I don't normally get stopped out by the stop I have given to the broker. As you probably know, "stops" is a whole can of worms to which there is a variety of approaches depending on one's financial constitution. Far stops are safe from the noise & spikes of random price fluctuation and incur larger risk provision in your money management, while near stops minimise risk (and thus allow greater gearing within one's risk tolerance) but are more likely to be taken out by the aforementioned spikes etc.
With SB-ing I use their charts/prices to gauge where I want my stops seeing as it's their game and no point in using someone else's rules. Note stops - since I use two. Firstly, a far one which I lodge with broker and this is well out of spike and stupidity range, and at a point where the trade has definitely failed and would in normal circumstances haves been picked up earlier. In fact it's really an emergency to keep you out of bankruptcy provision.
My second stop is private to me - a mental one - and is usually around or just outside support or resistance where again, the odds are that the trade's no good. This is the hard bit - the temptation is to hang on when this mental stop is approached or reached: "it might just get better" - totally the wrong mental attitude. I've learned the hard way that you note this stop on your log when you open the trade and when you reach it you exit - no ifs or buts. Sometimes this may not be to your advantage (hindsight's wonderful!) but on balance it saves one's bacon to trade another day - market's always there. Once you get in the discipline of acting on the mental stop it becomes a reflex action. I understand that you design aircraft wings, well those that fly them have their pre-planned actions to deal with emergencies and are trained to act reflexively for any standard situation - as is reaching your stop. As a trader I believe we should act the same way and I think there is a lot of similarity between flying and trading in the mental approach required.
Of course, the mental stop means monitoring the trade and this isn't always possible: sometimes I can't be doing with glued to the screen all day, so the "long stop" just has to be played and my money management is allied to that. Thus you can do a brief trade - using the near stop or if you have other things to do use the far one.
I now use my stop as a way of judging the quality of my entry (I know perceived wisdom is that exits are more important, but I find good entries help me to make good trades). In post trade analysis I calculate MAE (maximum adverse excursion - ie how much did the trade go against me) as a % of the stop distance. I found these data quite useful in getting my broker stop as safely near to my entry as possible. For swing trades I have seen books reccommend 2-3 X ATR as safe distance (saw some calculation that the 2 sigma level was about 2.5xATR). For my style of trading I found that using 1 xATR, my MAE is usually in the range 20-40%, [except for those lovely trades that go straight into profit] getting closer to 20% as a norm. (Those wonderful spreadsheets can tell you so much!) I'm still collecting data on this but it will allow me eventually to get my stops narrower and thus get max safe gearing on the average trade. This is possibly an exercise that would be illuminating for any account as it's specific to one's personal style.
But to get back to slippage etc. I note you are concerned with volume & spread - on a practical level these haven't (apparently) affected me. I suspect they may become important for the big boys but for us minnows who don't suffer their restrictions, life I suspect is much easier. Under normal conditions when I open/close a trade I get the price on the ticket - but I'm small beer, don't scalp and have more important things to worry about - like is the trade any good! On broker activated stops - again under normal conditions it's either spot on or pretty close - nothing that would cause me to start a thread on T2W. I have on just a few occasions experienced proper slippage when the market or the instrument is very busy - typically this is around open when the price just goes straight through your stop level. But I've changed my ways since then (don't berate the broker's nasty little habits - treat him like an obnoxious work colleague and find ways to cope) so that it's not a real problem. Nothing can cope with the domesday scenario - just like flying, all 4 engines can fail and that's not really covered in the manual.
Hope this (very personally-biased pov) is of some help. I wonder if it might be just as effective to knock off say, a nominal 5% from the results after using unfactored inputs. Since you are doing the full maths monty anyway, it could be an interesting comparison. Keep up the good work.:smart:
Last edited:
Just to add to this, IGIndex also have two useful features with regards stops and mental stops:
1) I believe that IGIndex "Stop" is based on the "mid-price", which helps a bit with opening spikes caused by wide spreads etc..where as other SB companies use the corresponding spread bid or ask price.
2) IGIndex have an "Alerts" feature which you can use to set against your "mental" stop, to email or SMS you of a mental price being hit.
megamuel
Experienced member
- Messages
- 1,098
- Likes
- 164
It's threads like these that make me feel really stupid! 😕 However, the bits I can understand I have found very interesting! I like your approach Sal although it is not technical analysis as I know it. You seem to be looking at statistics/probabilities as opposed to chart patterns, support and resistance, price history etc... Do you intend on looking at these things at some point? Any idea on how you will decide to enter a trade yet?
Talking about stats and probabilities, I remember once I thought I'd discovered a holy grail. I thought, I'll enter a trade at £1 per point with a 50pt stop loss and 50pt take profit. If I win, great, £50! If I lose, double my stake to £2 per point, and do the same again. Then I discovered that this betting system had a name (martingale) and it was destined to fail! Actually it can work, but you'd need such deep pockets that the reward is tiny compared to what you are risking. Actually I must admit, I am still kind of intrigued by the idea of martingale/mean reversion/and scaling in combined together... But I'm just not clever enough to do it. Maybe you could do some spreadsheets and add some partial differentiation wizardry to it 😉 (I have no idea what that means!).
Anyway, its nearly 4am and I think my sleeping pill is kicking in so this probably makes no sense at all but just thought I'd show my support and wish you good luck. In the future though can you please try and dumb things down a bit for me though?! 😆
All the best,
Sam.
Talking about stats and probabilities, I remember once I thought I'd discovered a holy grail. I thought, I'll enter a trade at £1 per point with a 50pt stop loss and 50pt take profit. If I win, great, £50! If I lose, double my stake to £2 per point, and do the same again. Then I discovered that this betting system had a name (martingale) and it was destined to fail! Actually it can work, but you'd need such deep pockets that the reward is tiny compared to what you are risking. Actually I must admit, I am still kind of intrigued by the idea of martingale/mean reversion/and scaling in combined together... But I'm just not clever enough to do it. Maybe you could do some spreadsheets and add some partial differentiation wizardry to it 😉 (I have no idea what that means!).
Anyway, its nearly 4am and I think my sleeping pill is kicking in so this probably makes no sense at all but just thought I'd show my support and wish you good luck. In the future though can you please try and dumb things down a bit for me though?! 😆
All the best,
Sam.
Cable Sal
Member
- Messages
- 70
- Likes
- 7
Hi Sam,
Lol, you made me laugh. The funny thing is that a lot of the maths only makes sense to me after midnight, not during the day.
Yes, I'll be looking into areas that are more pure technical analysis and price action as things develop here.
What I'm doing at the moment is building myself a simulator, one that's fully within my control and built to my specification, so that I can test different TA methodologies, indicators, etc, to see what works best for me and my style. It's not what I particularly set out to do, it's just developed from my initial goal of "pulling it apart to see how it works".
Some of what I've done and will go on to do wouldn't be necessary if I already had a SB account, but I know for sure if I open one I'll get distracted from my plan. It's good learning anyway to try and mimic how a SB account would behave, it makes me ask questions that I might not otherwise have thought about until they were costing me money.
First of all I needed a small sample of an instrument that I could feed my simulator - I chose stocks because I understand them without having to see how a SB platform functions. I specifically picked stocks that I'd not heard of so that I wouldn't be influenced by any of their fundamentals. I've been recording the bid, ask and volume of 11 stocks twice a day, and plotting those three figures on graphs in a spreadsheet. Over time, those graphs will fill with data that will become my charts to analyse.
On a separate sheet I have my SB calculator. At the moment I can enter up to 5 numbers on this sheet, an opening ask and bid; the bet size; and a closing ask and bid.
If I punch in just an opening ask and bid, it'll tell me what the SB guaranteed stop positions are (I'm working on replacing this with something more sophisticated), one stop for long and one stop for short. It gives me two breakeven prices (long and short), two prices where I'd breakeven if I scaled out half my bet, two prices if I wanted to keep all my bet on until it achieved 2:1 reward : risk. It calculates everything at twice minimum bet size. The prices that I give it are the ones I get from Yahoo, it gives me back the answers in Yahoo prices too, so that I can look for those price triggers on my graphs. The calculations in the background take account of the spreads.
If I then put in a closing ask and bid it'll tell me what my profit / loss was compared to the opening price. At this point it'll ask me to tell it what bet size I'm talking about, so I can calculate what profit / loss I've made on half the bet.
So what I do with this is I keep a running total on my graphs sheet. I work out my running profits / losses, assuming that I went in either long or short on a strict set of repeatable criteria. It allows me to see what would have happened over time and compare:
a) leaving everything in until a strict set of repeatable criteria dictates a change in trend, then pulling everything out;
b) scaling out half at breakeven, then pulling the rest out when a) above is seen;
c) leaving everything in until 2:1 reward:risk is seen, then pulling it all out;
d) the stop overrides everything, immediately pulling everything out recording a loss.
At the moment I've got two crazy things going on that need serious attention: my entry and exit points.
I'm paying attention to my stop placement first, because that affects the breakeven calculations and the 2:1 risk reward calculations. I'm going to replace the guaranteed stop with a manual stop (which will simulate the mental stop discussed in an earlier post, my operational with-losses exit). I can compare different methodologies of stop finding, and see what that does to my profit/loss profile for a), b), c) and d) above.
This is where TA starts to come in.
Once I've got my stop placement worked out I'll have a look at entry methods then and see if I can improve profitability. The "strict set of repeatable criteria" that I use at the moment are laughable, but very consistant and easy to apply. I can't fool myself when I backtrack into thinking that I would have done something different. Judging support and resistance will be more difficult to model, and tricky not to fool myself. Chart pattern analysis is easier, along with some of the indicators, but support and resistance are so important I won't ignore them just because I can't write a formula to model them.
I might not be able to back test, but I will be able to use everything that I've built to this point to see how my ability to judge support and resistance affects my profitability. Anyway, I'm not thinking too much about that just yet, methodology may come along in the meantime.
Gosh, explaining my thought process has inspired me to get back to the spreadsheets.
See you later,
Sal
Lol, you made me laugh. The funny thing is that a lot of the maths only makes sense to me after midnight, not during the day.
Yes, I'll be looking into areas that are more pure technical analysis and price action as things develop here.
What I'm doing at the moment is building myself a simulator, one that's fully within my control and built to my specification, so that I can test different TA methodologies, indicators, etc, to see what works best for me and my style. It's not what I particularly set out to do, it's just developed from my initial goal of "pulling it apart to see how it works".
Some of what I've done and will go on to do wouldn't be necessary if I already had a SB account, but I know for sure if I open one I'll get distracted from my plan. It's good learning anyway to try and mimic how a SB account would behave, it makes me ask questions that I might not otherwise have thought about until they were costing me money.
First of all I needed a small sample of an instrument that I could feed my simulator - I chose stocks because I understand them without having to see how a SB platform functions. I specifically picked stocks that I'd not heard of so that I wouldn't be influenced by any of their fundamentals. I've been recording the bid, ask and volume of 11 stocks twice a day, and plotting those three figures on graphs in a spreadsheet. Over time, those graphs will fill with data that will become my charts to analyse.
On a separate sheet I have my SB calculator. At the moment I can enter up to 5 numbers on this sheet, an opening ask and bid; the bet size; and a closing ask and bid.
If I punch in just an opening ask and bid, it'll tell me what the SB guaranteed stop positions are (I'm working on replacing this with something more sophisticated), one stop for long and one stop for short. It gives me two breakeven prices (long and short), two prices where I'd breakeven if I scaled out half my bet, two prices if I wanted to keep all my bet on until it achieved 2:1 reward : risk. It calculates everything at twice minimum bet size. The prices that I give it are the ones I get from Yahoo, it gives me back the answers in Yahoo prices too, so that I can look for those price triggers on my graphs. The calculations in the background take account of the spreads.
If I then put in a closing ask and bid it'll tell me what my profit / loss was compared to the opening price. At this point it'll ask me to tell it what bet size I'm talking about, so I can calculate what profit / loss I've made on half the bet.
So what I do with this is I keep a running total on my graphs sheet. I work out my running profits / losses, assuming that I went in either long or short on a strict set of repeatable criteria. It allows me to see what would have happened over time and compare:
a) leaving everything in until a strict set of repeatable criteria dictates a change in trend, then pulling everything out;
b) scaling out half at breakeven, then pulling the rest out when a) above is seen;
c) leaving everything in until 2:1 reward:risk is seen, then pulling it all out;
d) the stop overrides everything, immediately pulling everything out recording a loss.
At the moment I've got two crazy things going on that need serious attention: my entry and exit points.
I'm paying attention to my stop placement first, because that affects the breakeven calculations and the 2:1 risk reward calculations. I'm going to replace the guaranteed stop with a manual stop (which will simulate the mental stop discussed in an earlier post, my operational with-losses exit). I can compare different methodologies of stop finding, and see what that does to my profit/loss profile for a), b), c) and d) above.
This is where TA starts to come in.
Once I've got my stop placement worked out I'll have a look at entry methods then and see if I can improve profitability. The "strict set of repeatable criteria" that I use at the moment are laughable, but very consistant and easy to apply. I can't fool myself when I backtrack into thinking that I would have done something different. Judging support and resistance will be more difficult to model, and tricky not to fool myself. Chart pattern analysis is easier, along with some of the indicators, but support and resistance are so important I won't ignore them just because I can't write a formula to model them.
I might not be able to back test, but I will be able to use everything that I've built to this point to see how my ability to judge support and resistance affects my profitability. Anyway, I'm not thinking too much about that just yet, methodology may come along in the meantime.
Gosh, explaining my thought process has inspired me to get back to the spreadsheets.
See you later,
Sal
alan5616
Established member
- Messages
- 927
- Likes
- 198
Congratulations, Cable Sal, for putting such an intelligent and well written journal, on display. It makes absorbing reading. Not yer average Rookie-type missives; that's for sure 🙂
If you decide to venture into the world of forex, 0007 has made some very good points. I trade forex full time and the biggest mistake a lot of newbies make is to have a stop that is too tight. Trades need room to breathe. I dread to think of the number of losses I incurred by trying to trade with a 15-20 pip stop, before I ditched this idea. The slightest retracement results in the position being closed.
Good luck with whichever instrument(s) you end up trading.
If you decide to venture into the world of forex, 0007 has made some very good points. I trade forex full time and the biggest mistake a lot of newbies make is to have a stop that is too tight. Trades need room to breathe. I dread to think of the number of losses I incurred by trying to trade with a 15-20 pip stop, before I ditched this idea. The slightest retracement results in the position being closed.
Good luck with whichever instrument(s) you end up trading.
Cable Sal
Member
- Messages
- 70
- Likes
- 7
Thanks Alan, I appreciate the encouragement.
ATR(14)
So, I've removed the guaranteed stop out of my model and I'm about to find out what happens when I use the 14 day average true range value as a stop gap. This would represent my mental stop rather than broker stop - to use 0007's terminology.
I'm doing a bit of a fudge at the moment, but not an important one as far the experiment is concerned. I have less than three weeks price info, and I haven't been collecting the high and low prices.
Fudge 1: To calculate the ATR(1), I'm using the current midday price and the current day close, and saying whichever is high was the highest and whichever is low was the lowest. This means that my ATR(1) isn't as wide as reality, so a stop based on it would be more prone to being caught out by normal spikey price actions. I do collect the close of the previous day, so I'm able to accurately throw that into the mix. It's not terribly important because I'm equally not collecting the high and low points of spikes that would trigger a stop in my data - so the fudge is cancelling itself. maybe I should call it the average fudged range. :cheesy:
Fudge 2: From day 15 onwards in my price collecting, I'm able to calculate a fudged ATR(14), but prior to that I don't have 14 days of data to average across. To give me something to work with early in the time series with I'm taking an accumulating average, averaging across first 2, then 3, then 4 days, etc, until I reach 14, then I revert back to moving 14 day averages. It's a bit of a fudge, but I'm used to this one, I use it in the day job now and then. As time goes on, the first fudged 14 days become irrelevant.
Anyway, the results are looking significant, so I'm gonna get back to the spreadsheet.
See y'all later,
Sal 👍
ATR(14)
So, I've removed the guaranteed stop out of my model and I'm about to find out what happens when I use the 14 day average true range value as a stop gap. This would represent my mental stop rather than broker stop - to use 0007's terminology.
I'm doing a bit of a fudge at the moment, but not an important one as far the experiment is concerned. I have less than three weeks price info, and I haven't been collecting the high and low prices.
Fudge 1: To calculate the ATR(1), I'm using the current midday price and the current day close, and saying whichever is high was the highest and whichever is low was the lowest. This means that my ATR(1) isn't as wide as reality, so a stop based on it would be more prone to being caught out by normal spikey price actions. I do collect the close of the previous day, so I'm able to accurately throw that into the mix. It's not terribly important because I'm equally not collecting the high and low points of spikes that would trigger a stop in my data - so the fudge is cancelling itself. maybe I should call it the average fudged range. :cheesy:
Fudge 2: From day 15 onwards in my price collecting, I'm able to calculate a fudged ATR(14), but prior to that I don't have 14 days of data to average across. To give me something to work with early in the time series with I'm taking an accumulating average, averaging across first 2, then 3, then 4 days, etc, until I reach 14, then I revert back to moving 14 day averages. It's a bit of a fudge, but I'm used to this one, I use it in the day job now and then. As time goes on, the first fudged 14 days become irrelevant.
Anyway, the results are looking significant, so I'm gonna get back to the spreadsheet.
See y'all later,
Sal 👍
Cable Sal
Member
- Messages
- 70
- Likes
- 7
First fix
Okaaaaa-ay! Looking good here.
What I had before (SB version 1):
A guaranteed stop which was set a fixed 7.5% away, costing me a premium of 0.7% extra width on the spread.
A comedy set of strict repeatable entry criteria, which have the unfortunate effect of often making me buy at the peak and sell in the trough.
A set of exit criteria which enabled comparison between the following:
a) leaving everything in until a strict set of repeatable criteria dictates a change in trend, then pulling everything out;
b) scaling out half at breakeven, then pulling the rest out when a) above is seen;
c) leaving everything in until 2:1 reward:risk is seen, then pulling it all out;
d) the stop overrides everything, immediately pulling everything out recording a loss.
What I've got now (SB version 2):
A manual non-guaranteed stop which is placed a distance away roughly based on the stock's ATR(1) to ATR(14), a bit fudged, but consistently so.
The spread no longer has the premium 0.7%, but there's now a negative financial impact of 5% net on all stopped trades to account for slippage risk.
A comedy set of strict repeatable entry criteria, which have the unfortunate effect of often making me buy at the peak and sell in the trough.
A set of exit criteria which enables comparison between the following:
a) scaling out half at breakeven, then pulling the rest out when a strict set of repeatable criteria dictates a change in trend;
b) leaving everything in until 2:1 reward:risk is seen, then pulling it all out;
c) the stop overrides everything, immediately pulling everything out recording a loss.
The result:
Unsurprisingly, I wasn't showing a running total profit with what I had before, but I was reasonably pleased that my risk management was working, the crazy entry and exit criteria weren't being allowed to run up enormous loss. More than anything, I was really, really pleased that the model was working. Applying the methodology to my tracked prices, starting 27/07/10 up to 12/08/10, my losses were -£972.
I'm still not showing a decent profit, but I'm not concerned about that just yet, it's all moving in the right direction. My running total is now showing a loss of -£178 when I use the with-profits exit of "all out at 2:1 reward:risk", and a small profit of £36.97 when I use the with-profits exit of "scale-out half at breakeven, and take the other half at stop or when some crazy set of strict, repeatable criteria says the trend has changed".
Next fix
I'm not finished looking at stop placement, risk management or exit criteria, I will come back, maybe to compare different methodologies, definitely to look at the % risk to my account of some of the higher value shares. But for now, I'm gonna leave these alone, they're no longer my biggest problems.
The entry criteria that I'm using has become the biggest problem now, the biggest hit on profits / addition to losses. It doesn't make sense to tweak around, finessing other things while my entry criteria is resulting in me buying at peaks and selling at troughs so often.
There's a whole world of options out there, some of which are easier to model than others. A difficult one is chart analysis, I'm a good pattern analyst, so candlesticks particularly appeals. I know I could write formulae to turn candlestick charts and patterns into algorithms, but that'd take ages, I think I'll have to leave it alone for now and stick to stuff that's easy to convert into formulae.
What I'm looking for now is a quick and dirty solution, something that's easy to model and results in me mostly not going long at the high point and short at the low point. Time to research indicators....
Lol, I'm still just chuffed that I've turned out a positive figure. 😀 I'm feeling £36.97 richer, and there's still a fair few half positions running free at profit. It doesn't matter that it's virtual, doesn't matter that it's peanuts, it's like I've just beaten level 1 on a computer game, now for level 2.
By the way, does anybody know, when you're calculating ATR(14) is that 14 working days or 14 calendar days? I guess I should go look that up while I'm browsing for suitable indicators.
Bye for now
Happy Sal
😛arty:
Okaaaaa-ay! Looking good here.
What I had before (SB version 1):
A guaranteed stop which was set a fixed 7.5% away, costing me a premium of 0.7% extra width on the spread.
A comedy set of strict repeatable entry criteria, which have the unfortunate effect of often making me buy at the peak and sell in the trough.
A set of exit criteria which enabled comparison between the following:
a) leaving everything in until a strict set of repeatable criteria dictates a change in trend, then pulling everything out;
b) scaling out half at breakeven, then pulling the rest out when a) above is seen;
c) leaving everything in until 2:1 reward:risk is seen, then pulling it all out;
d) the stop overrides everything, immediately pulling everything out recording a loss.
What I've got now (SB version 2):
A manual non-guaranteed stop which is placed a distance away roughly based on the stock's ATR(1) to ATR(14), a bit fudged, but consistently so.
The spread no longer has the premium 0.7%, but there's now a negative financial impact of 5% net on all stopped trades to account for slippage risk.
A comedy set of strict repeatable entry criteria, which have the unfortunate effect of often making me buy at the peak and sell in the trough.
A set of exit criteria which enables comparison between the following:
a) scaling out half at breakeven, then pulling the rest out when a strict set of repeatable criteria dictates a change in trend;
b) leaving everything in until 2:1 reward:risk is seen, then pulling it all out;
c) the stop overrides everything, immediately pulling everything out recording a loss.
The result:
Unsurprisingly, I wasn't showing a running total profit with what I had before, but I was reasonably pleased that my risk management was working, the crazy entry and exit criteria weren't being allowed to run up enormous loss. More than anything, I was really, really pleased that the model was working. Applying the methodology to my tracked prices, starting 27/07/10 up to 12/08/10, my losses were -£972.
I'm still not showing a decent profit, but I'm not concerned about that just yet, it's all moving in the right direction. My running total is now showing a loss of -£178 when I use the with-profits exit of "all out at 2:1 reward:risk", and a small profit of £36.97 when I use the with-profits exit of "scale-out half at breakeven, and take the other half at stop or when some crazy set of strict, repeatable criteria says the trend has changed".
Next fix
I'm not finished looking at stop placement, risk management or exit criteria, I will come back, maybe to compare different methodologies, definitely to look at the % risk to my account of some of the higher value shares. But for now, I'm gonna leave these alone, they're no longer my biggest problems.
The entry criteria that I'm using has become the biggest problem now, the biggest hit on profits / addition to losses. It doesn't make sense to tweak around, finessing other things while my entry criteria is resulting in me buying at peaks and selling at troughs so often.
There's a whole world of options out there, some of which are easier to model than others. A difficult one is chart analysis, I'm a good pattern analyst, so candlesticks particularly appeals. I know I could write formulae to turn candlestick charts and patterns into algorithms, but that'd take ages, I think I'll have to leave it alone for now and stick to stuff that's easy to convert into formulae.
What I'm looking for now is a quick and dirty solution, something that's easy to model and results in me mostly not going long at the high point and short at the low point. Time to research indicators....
Lol, I'm still just chuffed that I've turned out a positive figure. 😀 I'm feeling £36.97 richer, and there's still a fair few half positions running free at profit. It doesn't matter that it's virtual, doesn't matter that it's peanuts, it's like I've just beaten level 1 on a computer game, now for level 2.
By the way, does anybody know, when you're calculating ATR(14) is that 14 working days or 14 calendar days? I guess I should go look that up while I'm browsing for suitable indicators.
Bye for now
Happy Sal
😛arty:
Cable Sal
Member
- Messages
- 70
- Likes
- 7
Gosh, it's been interesting browsing the forums looking at different opinions on indicators, just a mention of indicators seems to raise the blood pressure of some traders. I never imagined that a moving average could bother people that much. I spend half my time in my day job trying to get people het up about indicators, lol.
Acknowledging the risk of starting a blazing row between the pro and the anti camps, I'd like to put my current thought train on the record. I'll try to do this as diplomatically as possible - please don't fight over this one people - you'll get me banned for flaming or trolling or whatever the appropriate label is.
I'm currently sat in my day job office (writing my journal so blatantly not working, although I should be). Yes it is Sunday, yes I am doing overtime, no I'm not getting paid for it, yes I will go AWOL at some point to get the hours back. I'm here because I'm getting a new suite of Key Performance Indicators that I developed loaded onto the company business intelligence suite. They're a mix of familiar moving averages, comparators and indicies - in fact I've robbed some techniques from TA to improve some of my day job indicators.
When making decisions in trading there are comparatively few options: choose your instrument, your platform, your style, your timeframe, go long or short, when to enter, when to exit, where to place stops and limits, etc. There's a single, clear goal: put money in your account, don't lose it. There's a chart for everything that you're trading that you can pip-watch and take immediate decisions on that will instaneously (nearly) result in a measureable impact on your account. The information is available and clear, decision making is difficult because it's impactful.
It's different in business, where chaos theory prevails. The flap of a butterfly's wings in the marketing department in May can create havoc in the finance department in August. There's no single chart to follow to tell us if things are going right or wrong, we can't follow the price action and take a single decision that will immediately determine our profitability. So, we write a business plan, defining what our goals are - some of which will be financial, some won't, we list the actions that we're going to take that we think are going to enable us to achieve our goals and then we monitor. We monitor whether the actions are going ahead on budget and to schedule, and we monitor the impact that the actions are having to decide whether we're on track, or if we need to take different action to what we thought. It's monitoring the impact of actions that drives the art / science of Key Performance Indicators.
Example: we're being sued by members of the general public for slips, trips and falls in our public areas - one of the financial indicators (variance against budget for claims) is well into the red. What do we do?
Answer A: improve how we challenge claims to bring down the costs;
Answer B: investigate the costs of improving surfaces at hot-spots;
Answer C: bring in more staff to improve the daily inspection regime; or
Answer D: something else.
The answer to this one was in the indicator "number of slips, trips and falls per million visitors", presented as a simple moving average to reduce the spikiness. The result was actually a downward trend - we're getting less accidents per visitor than ever before - it's just that our visitor numbers are soaring (yeay!). Benchmarking the result against other organisations we find that we're actually doing really, really well, but something's not right about our claims bill. Create another indicator: "average claim cost per slip/trip/fall" - year on year it's been going up. So, the answer was A: improve how we challenge claims, because after a bit more stats analysis we discovered that we were being targeted by ambulance chasers and fraudulent claimants, we were a bit of an easy touch. I'll start assessing the results of the procedural changes in November, which is when the new actions have all been finalised and implemented.
It's a bit different to designing aircraft wings, but the maths is sometimes similar. It's nothing at all like trading, except that the statisical treatment of historical data is sometimes similar.
Indicators have a place in my plan at the moment. Trading indicators are summaries of the price action, a way to mathematically present trends. I can use boolean algebra to combine several indicators into a decision making model: if x is positive and y is negative then sell else hold. Given a lot of learning and practice and full-time sitting in front of live charts, I expect following the price action to be more profitable than decision making by indicators, but I can't model that, and it's incredibly difficult to backtest theories.
How to wipe the knowledge of what comes next in order to assess my ability to make decisions on historical price action? The same with chart pattern analysis, it's very easy to spot the signals in the chart when you know what happened next, but could I spot them without that knowledge?
What I need is to have a testable trading plan, something that I know will produce a positive expectancy. Something that I can take live and run with. Once I'm comfortable with trading my trading plan, I can live-trade my indicators and formulae-derived conditions, while paper trading pattern analysis and price action. I can forward test for as long as it takes to learn the skills that I need to be more profitable with the hard to model stuff than I am with the easy to model stuff.
The opinion that I'm forming so far through my research is that there's no right answers to the tools and methods and decisions in trading. They're all right for some people some of the time. Losses come from applying a right thing in the wrong way, or at the wrong time, or interpreting the result into the wrong action. This is equally true of price action, pattern analysis, indicators, fundamentals, psychology, everything.
Anyway, this business intelligence system isn't going to programme itself - I'd best get back to the day job.
Sal
Acknowledging the risk of starting a blazing row between the pro and the anti camps, I'd like to put my current thought train on the record. I'll try to do this as diplomatically as possible - please don't fight over this one people - you'll get me banned for flaming or trolling or whatever the appropriate label is.
I'm currently sat in my day job office (writing my journal so blatantly not working, although I should be). Yes it is Sunday, yes I am doing overtime, no I'm not getting paid for it, yes I will go AWOL at some point to get the hours back. I'm here because I'm getting a new suite of Key Performance Indicators that I developed loaded onto the company business intelligence suite. They're a mix of familiar moving averages, comparators and indicies - in fact I've robbed some techniques from TA to improve some of my day job indicators.
When making decisions in trading there are comparatively few options: choose your instrument, your platform, your style, your timeframe, go long or short, when to enter, when to exit, where to place stops and limits, etc. There's a single, clear goal: put money in your account, don't lose it. There's a chart for everything that you're trading that you can pip-watch and take immediate decisions on that will instaneously (nearly) result in a measureable impact on your account. The information is available and clear, decision making is difficult because it's impactful.
It's different in business, where chaos theory prevails. The flap of a butterfly's wings in the marketing department in May can create havoc in the finance department in August. There's no single chart to follow to tell us if things are going right or wrong, we can't follow the price action and take a single decision that will immediately determine our profitability. So, we write a business plan, defining what our goals are - some of which will be financial, some won't, we list the actions that we're going to take that we think are going to enable us to achieve our goals and then we monitor. We monitor whether the actions are going ahead on budget and to schedule, and we monitor the impact that the actions are having to decide whether we're on track, or if we need to take different action to what we thought. It's monitoring the impact of actions that drives the art / science of Key Performance Indicators.
Example: we're being sued by members of the general public for slips, trips and falls in our public areas - one of the financial indicators (variance against budget for claims) is well into the red. What do we do?
Answer A: improve how we challenge claims to bring down the costs;
Answer B: investigate the costs of improving surfaces at hot-spots;
Answer C: bring in more staff to improve the daily inspection regime; or
Answer D: something else.
The answer to this one was in the indicator "number of slips, trips and falls per million visitors", presented as a simple moving average to reduce the spikiness. The result was actually a downward trend - we're getting less accidents per visitor than ever before - it's just that our visitor numbers are soaring (yeay!). Benchmarking the result against other organisations we find that we're actually doing really, really well, but something's not right about our claims bill. Create another indicator: "average claim cost per slip/trip/fall" - year on year it's been going up. So, the answer was A: improve how we challenge claims, because after a bit more stats analysis we discovered that we were being targeted by ambulance chasers and fraudulent claimants, we were a bit of an easy touch. I'll start assessing the results of the procedural changes in November, which is when the new actions have all been finalised and implemented.
It's a bit different to designing aircraft wings, but the maths is sometimes similar. It's nothing at all like trading, except that the statisical treatment of historical data is sometimes similar.
Indicators have a place in my plan at the moment. Trading indicators are summaries of the price action, a way to mathematically present trends. I can use boolean algebra to combine several indicators into a decision making model: if x is positive and y is negative then sell else hold. Given a lot of learning and practice and full-time sitting in front of live charts, I expect following the price action to be more profitable than decision making by indicators, but I can't model that, and it's incredibly difficult to backtest theories.
How to wipe the knowledge of what comes next in order to assess my ability to make decisions on historical price action? The same with chart pattern analysis, it's very easy to spot the signals in the chart when you know what happened next, but could I spot them without that knowledge?
What I need is to have a testable trading plan, something that I know will produce a positive expectancy. Something that I can take live and run with. Once I'm comfortable with trading my trading plan, I can live-trade my indicators and formulae-derived conditions, while paper trading pattern analysis and price action. I can forward test for as long as it takes to learn the skills that I need to be more profitable with the hard to model stuff than I am with the easy to model stuff.
The opinion that I'm forming so far through my research is that there's no right answers to the tools and methods and decisions in trading. They're all right for some people some of the time. Losses come from applying a right thing in the wrong way, or at the wrong time, or interpreting the result into the wrong action. This is equally true of price action, pattern analysis, indicators, fundamentals, psychology, everything.
Anyway, this business intelligence system isn't going to programme itself - I'd best get back to the day job.
Sal
DionysusToast
Legendary member
- Messages
- 5,965
- Likes
- 1,501
Hi Sal
Interesting thread.
How many permutations of the hundreds (thousands?) of TA 'methodologies' are you willing to test?
What specifically leads you to believe that these TA methodologies are actually worth investigating? Is there any evidence you have seen that they work - or is it mostly anecdotal?
Have you considered that a scientific approach to the problem you are trying to resolve (how to pull money from the markets) could just burn 2 years of your life yet bring you no closer to your goals? Have you considered any other 'less intelligent' ways of pulling $$$ from the markets?
Cheers
DT
alan5616
Established member
- Messages
- 927
- Likes
- 198
DT has raised a challenging question here. His observation, that you could waste 2 years of your life without getting any closer to your goals, hits the nail on the head.
If I have learnt anything in my trading life it's that markets change constantly. The strategy that worked, so well, last week, loses money the following week. I've lost too much money using trend following strategies that I'm now a confirmed counter trend trader. It works for me but, realise that it's not for everyone.
Cable Sal
Member
- Messages
- 70
- Likes
- 7
Hey DT,
Easy answers to these ones, just not necessarily in the order that you've asked them :cheesy:
I'm prepared to go as far as October playing around with TA to see if I can get something that looks like a developing trading plan out of it. If I can't, then I walk away from trading. If I've got something that looks promising, I work on it until December, when it's got to be ready for paper trading, if not, then I walk away from trading. I'm willing to be more flexible with myself and timescales beyond that point, I might decide to paper trade for longer than my original timescale - I don't want to push myself faster than I'm comfortable with.
How many permutations am I willing to try? Well, initially, as many as I need in order to create the outline of a trading plan in October. If I go on to trade, and continue to use indicators to inform my trading, and it's succesful, there's no reason why I wouldn't keep on trying new things forever. Why not test out some new theories alongside the method that I'm running live with?
Have I considered 'less intelligent' ways - well I made nearly £300k in position trading fundamentals, but it was a harem-scarum ride that I don't want to get back on. If I can't get TA to work for me, then it's back to the quiet life for me. I can get plenty of cash doing what I'm doing, and plenty of thrills playing computer games and just living my life, I don't need trading.
Why do I believe that TA will work for me? I'm attempting not to believe anything, but to test a new (new for me) way of pulling money out of the markets. I have never witnessed anybody else's trades first hand - so every single last thing is anecdotal. I have no proof of anything beyond the fact that markets exist and I've made money out of them in the past. TA sits comfortably with me as I have an engineers mind, I naturally want to pull things apart and see how they work, then build a model to see if I can simulate them and improve my design / decision making. I use TA everyday in a business envirnoment, where I have no other option but to analyse data to make decisions. It works here, so maybe it'll work for me in the markets too.
If I burn away 2 years of my life, but have fun doing it, then that'll have achieved one of my goals for the next 2 years - to have fun doing whatever I'm doing. If I'm not having fun I'll stop and go do something else instead.
Hope that clarifies?
Sal
