Well...It Is All Depens On Your Winning Percentage

[searchlight]

HI,
In this tread I will share with you, the system which I think can generate good, healthy debate and add to the knowledge already exist; even in the end, decided that it is no good at all.

This system works on the basis that one knows his/her winning percentage. So we will assume that (either extensive back testing or trading records) this data available.
If you ask the question ; What percentage of my trading capital should I risk for per trade? Answer come back as 1% or 2%. more often than not without reasoned explanation. Rather then static figure this system advocates dynamic figure according to ones risk profile.hence the title of the tread.
here it is:
first we need a statistical table which shows relation between a winning percentage and the 1% chance of longest loosing sequence to expect when trading.
LONGEST LOOSING SEQUENCE TO EXPECT***
WINNING PERCENTAGE 1% CHANCE
25% ---------------------------------17
26% ----------------------------------16
27% ---------------------------------15
28% ----------------------------------15
29% ----------------------------------14
30-31% -----------------------------13
32-34% ------------------------------12
35-36% ------------------------------11
37-39% ------------------------------10
40% ----------------------------------10
41-43% ------------------------------9
44-45% -------------------------------8
46-48%--------------------------------8
49-52%--------------------------------7
53% -----------------------------------7
54-60% ------------------------------6
61-63% -------------------------------5
64-68% -------------------------------5
69-77% -------------------------------4
78% ----------------------------------4
79-80% ------------------------------3
*** I have this table for several years now...I don't remember originally which publication I get it from. I always presumed those figures to be correct.

looking at this table we can see that someone who's winning percentage is 25% should prepare him/herself as many as 17 losing trades in a row, while someone who has 53% winning percentage need only worry about 7 loosing trades in a row.
On this figures , The 25% trader could afford to risk 5.88% of his/her initial trading capital on each trade(100/17), and the 53% winner could afford to risk 14.28% of their initial trading capital on each trade.(100/7). In each case, they would only have a 1% of going broke.
of course for the traders, making their living from trading 1% still unacceptable level of risk for blowing account.
to prevent this, to strategies suggested to implement;
1-risk only half the amounts indicated above
2-do not risk a set percentage of initial trading capital, instead risk percentage of present trading capital. If your trading capital grows, so size of the capital you risk, If you have lost money , size of the capital you risked will be correspondingly smaller.

So what do you think? Is it worth considering?

Best regards,

 

[TheBramble]

When you consider statistical averages you also need to consider he outliers that comprise the fat (very fat) tails of normal distribution in this example. Working with the data you provide perhaps you’d care to calculate what percentage of your trading capital you could, statistically, trade on a system with 99.99% probability of success?

I’ve had enough experience in outlierland to appreciate the commonsense of minimal risk exposure. Of course, there are those I imagine that play a full hand and are absolutely maximising their gains on every trade and have been doing so for years, but I don’t just know any of them personally.

You say more often than not 1% or 2% comes back as an answer to the question of max capital risk per trade without reasoned explanation. I think the fact it comes back so often is because that may be where many traders who have been trading for a long time end up. There’s probably a reason for that too. Just looking at my most recent 6 months performance I have had two runs of consecutive losers which fall way outside the ranges of the data you provide. I don’t think it pays to get too clever, but then again, I am rather unadventurous and boring.

 

[damianoakley]

HI,
In this tread I will share with you, the system which I think can generate good, healthy debate and add to the knowledge already exist; even in the end, decided that it is no good at all.

This system works on the basis that one knows his/her winning percentage. So we will assume that (either extensive back testing or trading records) this data available.
If you ask the question ; What percentage of my trading capital should I risk for per trade? Answer come back as 1% or 2%. more often than not without reasoned explanation. Rather then static figure this system advocates dynamic figure according to ones risk profile.hence the title of the tread.
here it is:
first we need a statistical table which shows relation between a winning percentage and the 1% chance of longest loosing sequence to expect when trading.
LONGEST LOOSING SEQUENCE TO EXPECT***
WINNING PERCENTAGE 1% CHANCE
25% ---------------------------------17
26% ----------------------------------16
27% ---------------------------------15
28% ----------------------------------15
29% ----------------------------------14
30-31% -----------------------------13
32-34% ------------------------------12
35-36% ------------------------------11
37-39% ------------------------------10
40% ----------------------------------10
41-43% ------------------------------9
44-45% -------------------------------8
46-48%--------------------------------8
49-52%--------------------------------7
53% -----------------------------------7
54-60% ------------------------------6
61-63% -------------------------------5
64-68% -------------------------------5
69-77% -------------------------------4
78% ----------------------------------4
79-80% ------------------------------3
*** I have this table for several years now...I don't remember originally which publication I get it from. I always presumed those figures to be correct.

looking at this table we can see that someone who's winning percentage is 25% should prepare him/herself as many as 17 losing trades in a row, while someone who has 53% winning percentage need only worry about 7 loosing trades in a row.
On this figures , The 25% trader could afford to risk 5.88% of his/her initial trading capital on each trade(100/17), and the 53% winner could afford to risk 14.28% of their initial trading capital on each trade.(100/7). In each case, they would only have a 1% of going broke.
of course for the traders, making their living from trading 1% still unacceptable level of risk for blowing account.
to prevent this, to strategies suggested to implement;
1-risk only half the amounts indicated above
2-do not risk a set percentage of initial trading capital, instead risk percentage of present trading capital. If your trading capital grows, so size of the capital you risk, If you have lost money , size of the capital you risked will be correspondingly smaller.

So what do you think? Is it worth considering?

Best regards,

Hi Searchlight,

Interesting, but it's not just about being able to survive the longest losing run, it's also about drawdown.

So, in your example above, you show that having a system with a 53% strike rate means that you can expect 7 losers in a row, which in turn means that you could risk 14%(!) on each trade. So, every time you suffered 3 or 4 losers in a row (which would happen quite frequently), then you'd lose half your capital! That would be one hell of a volatile account!

Whereas if you just risked 2% per trade, your 7 loser streak would cause a 14% drawdown, which would be a lot easier to deal with.


Thanks

Damian

 

[searchlight]

The Degrees,
I will follow the rules system advocates in order to answer your question.
99.99% winning ; I will take longest expected losing run as 1(smallest I can go). So he has 100 units to trade.
he risks 50 units (rule 1) ......... say he lost. he has left with 50 units.
he risks 25 units (rule 2).......... now he expect tens of trades before he sees a loosing run of 1. Even 4/5 winning trades will take him more then 250 units. What if he loses again?
Because we are talking 99.99% success rate here , as you can see system can cope with that as well.

on a lighter note though; If with the 99.99% success rate he loses 2 in a row, money he lost should be the least of his worries. He should enjoy the life while he can, kiss goodbye to his loved ones everyday. with his luck he won't be around for long...

OK. on a serious note, what percentage of your trading capital you risk if your success rate was 99.99% . If you say 2% fair enough. if it is more than that ,would you accept that ; ones success rate should play a role, for capital risked per trade. The question we have to find an answer is; what should be the correlation?

Damian,
Point taken. Because of the rules (risking % of present trading capital, instead of initial trading capital ) losses would be lower , but significant nevertheless. Gains also be equally significant. our quest is; is the system robust enough to produce a profit in the long run? personally, i dont know the answer to above question. I'm hoping to find out as the tread develops.
if this way of using success rate is to risky, and not worth it, Is success rate should play any role at all how the capital allocated for per trade.
think of this way;
2 traders ; equal trading capitals, same risk/reward ratio
one has 35% success rate, one has 70% success rate
should both risk 2% ?

best regards,

 

[damianoakley]

In my opinion no,

The trader with the 35% strike rate should risk 1% or less.



Damian

 

[Hoggums]

The Degrees,
now he expect tens of trades before he sees a loosing run of 1.

WRONG!!

You are implying your system has "memory" of when it has lost and can compensate by winning more after a series of losses.

If you have a system with 70% success rate and you have 3 losers in a row, then the chance that your next trade will be a winner is --- 70% !

 

[searchlight]

Damian,
the fact that, you suggesting the trader with the 35% success rate should risk 1% or less tells me you prepare the adjust capital risked per trade according to success rate.
Because everything else was the same. So what criteria you used to adjust the percentage downwards? Can we also use the same criteria to adjust percentage upwards?
thanks,

Hoggums,
I readily confess, I don't have in depth knowledge of statistics, only the basic one . I know it can be a very dangerous thing.
this is how i see it , any input to point me in the right direction greatly welcome.

Say i have 10% chance of having 6 losing trades in a row .It means If i trade 100 series of six trades each , ten of them would not contain a winning trade.Correct?

So if i have 1% chance of having 1 losing trade in a row , it means if i trade 100 series of one trade each, one of them would not contain a winning trade.Correct?

in the 10% example, after one set of 6 losing trade to expect a winner (although it could be argued that one nearer to the winning trade) wrong.Because there is a 9 set of six trades out there that contains no winning trades. they can be distributed in any order.

in the 1% example, after 1 losing trade. It looks like to me a certainty. I don't expect , I demand a winning trade. where am i going wrong?

Real reason i started this tread to compare with the help of members; fixed percentage of allocating capital for per trade versus one linked to the winning percentage . And see if any of them have a edge over the other. There seems to be no takers on the one linked to the winning percentage one. As The Degrees said, may be a good reason for it. would you care the comment?

thanks,

 

[Hoggums]

Damian,
the fact that, you suggesting the trader with the 35% success rate should risk 1% or less tells me you prepare the adjust capital risked per trade according to success rate.
Because everything else was the same. So what criteria you used to adjust the percentage downwards? Can we also use the same criteria to adjust percentage upwards?
thanks,

Hoggums,
I readily confess, I don't have in depth knowledge of statistics, only the basic one . I know it can be a very dangerous thing.
this is how i see it , any input to point me in the right direction greatly welcome.

Say i have 10% chance of having 6 losing trades in a row .It means If i trade 100 series of six trades each , ten of them would not contain a winning trade.Correct?

So if i have 1% chance of having 1 losing trade in a row , it means if i trade 100 series of one trade each, one of them would not contain a winning trade.Correct?

in the 10% example, after one set of 6 losing trade to expect a winner (although it could be argued that one nearer to the winning trade) wrong.Because there is a 9 set of six trades out there that contains no winning trades. they can be distributed in any order.

in the 1% example, after 1 losing trade. It looks like to me a certainty. I don't expect , I demand a winning trade. where am i going wrong?

Real reason i started this tread to compare with the help of members; fixed percentage of allocating capital for per trade versus one linked to the winning percentage . And see if any of them have a edge over the other. There seems to be no takers on the one linked to the winning percentage one. As The Degrees said, may be a good reason for it. would you care the comment?

thanks,

I think a 99% success rate is unrealistic - so lets deal with a far more likely scenario. 50% success. It's quite reasonable to trade a 50% success rate if your winners are larger than your losers. In fact the higher your profit factor the more likely is your win rate will be lower. But lets take 50% success. e.g. the toss of a coin.

You are saying this.....

Let's say you make 10 trades = and they are all losers. The probability that this happens is 0.5 *0.5 *....= 0.097% - unlikely - but it COULD happen. By your reckoning the 11th time you trade the probability it will win is 100%-0.097% = 99.902%

NONSENSE.

The next time you trade you have a 50% chance of winning - no matter what has happened before.

You forget that every time you trade you are starting from zero. The market doesn't remember what you've done. So if by some freak chance you've lost 20 trades in a row - the chances that you'll profit from the next 10 trades are EXACTLY THE SAME as the chances you would profit if you have won your last 20 trades.

 

[searchlight]

No, i am not saying that. What i am saying is that in your 50% example he starts with the base of 100 , after 10 losing trade, base drops the 90. He still has 50% chance of winning , out of base 90 , not 100. so he has a better chance of finding a winning trade
even though 50% remains the same.

 

[Rhody Trader]

Interesting, but it's not just about being able to survive the longest losing run, it's also about drawdown.

So, in your example above, you show that having a system with a 53% strike rate means that you can expect 7 losers in a row, which in turn means that you could risk 14%(!) on each trade. So, every time you suffered 3 or 4 losers in a row (which would happen quite frequently), then you'd lose half your capital! That would be one hell of a volatile account!

Whereas if you just risked 2% per trade, your 7 loser streak would cause a 14% drawdown, which would be a lot easier to deal with.

This is the key point in this whole discussion. The size of the drawdowns are what really matters. If you lose 50% of your money, you have to earn 100% just to get back to level. So when you try to determine position sizing based on statistics you have to look recovery times from the inevitable drawdowns, not just the chance of them happening.

Also, keep in mind that strings of bad results can happen in groups. For example, you could see a run of 5, followed by a good trade, followed by another run of 4. That sort of thing doesn't show up in the statistics when you just try looking at the odds of a run of 9 bad trades in a row.

 

[Tadragh1]

You forget that every time you trade you are starting from zero. The market doesn't remember what you've done. So if by some freak chance you've lost 20 trades in a row - the chances that you'll profit from the next 10 trades are EXACTLY THE SAME as the chances you would profit if you have won your last 20 trades.

Hoggums has a point here. As a side note - actually it seems, that even after heavy back-testing one cannot say that his probability of success is 50%. Quoting just one number assumes, that the outcomes of each position are i.i.d. variables and it is rather unlikely they are. So coming up with risk management rules based on just your sample probability of success may be an educating experience, though putting them into practice might prove dangerous.

 

[TheBramble]

These sorts of discussions can get bogged down in all sorts of issues of probability which can be fun, but are largely irrelevant when considering trading statistics. Trades are not coins. They don’t have binary outcomes. They have results ranging from total failure to complete success. The trading system gets modified in relation to the previous results. It therefore does have memory.

There is a branch of statistics with which you can work which will enable you to assign these factors and issues quantifiable weight within your calculations, but if you’re that deep you’re a statistician and not a trader. (Statisticians never get round to pulling the trigger).

The original intent of this thread was to examine the usefulness (or not) of working with your W:L as a statistical basis for establishing risk/position size. Those who caution against trying to squeeze the last drop out of probability out of possibility (and I am one of them too) are correct. Which is why, as I said in an earlier post here, most traders who have been trading a while come up with the same old boring 1% or 2% when questioned on risk size. They do so because that’s what has worked for them over time. It’s not quite so boring when you consider they probably went through the cleverer than clever phase of trying to be too cute with stats before coming to this conclusion. I know I did.

 

[Tadragh1]

Just for the sake of completeness of this educative thread - I am attaching an Excel table with probabilities of consecutive losses given a known, constant W/L ratio.
Caveat 1: As argued above, known, constant W/L ratios rather do not exist in the real world.
Caveat 2: When trying to make any money management rules based on such a table, you have to assume that the amount you win is equal to the amount you lose.
Caveat 3: Even if the amounts were even still such a table is pretty useless to use for money management, because consecutive losses are not the sole reasons for drawdowns and bankruptcy. Loss / Loss / Loss scenario is as bad as Win / Loss / Loss / Win / Loss / Loss / Win / Loss / Loss, if we assume WinAmount = LossAmount. Therefore you'd like actually to get a full drawdown distribution for money management purposes.

 

[JTrader]

Well...It Is All Depens On Your Winning Percentage

Well if you have 90% winners and you risk 3 times more than you stand to gain, the 90% winners stat does not mean a great deal in terms of ACTUAL PROFITS.