When to go live?

[CompoundTrader]

Hi trade2win, I've been involved with trading for several years and have been working full time on building a trading strategy since last October.

I have started live simulation trading this January, but the results are not good enough yet. I am looking for some guidance on how to use maths and statistics to identify when I am ready to go live.

Here is the last two weeks (date, net profit, gross profit, gross loss, cumulative net profit):

The statistics I am tracking so far:
-profit factor (ratio of gross loss to gross profit)
-ratio of net profit to maximum drawdown
-gross losses

The reason I am choosing these statistics is that I want to focus on keeping my losses as small as possible in order to use leverage effectively and compound my account. I believe if I prepare properly, total risk can be kept small, so I plan on starting with a $4,000 futures account and staking $10/tick.

Profit factor tells me how many dollars I earn for each dollar lost, and I would like this to be 3 or more. Also, in any rolling 5 day period, I want to have made at least 4x my maximum drawdown in that period. Both of these together will ensure that time to recovery after drawdown is as quick as possible.

Total gross losses are still far too high. There are various improvements I can make to my method to eliminate some of these losses. This will take time - the question is, when am I successful enough to begin live trading?

My question for trade2win is are there any other measurements I should be using to track how consistent I am, how much risk I am taking, and therefore what leverage to use? I have read about Kelly, but doesn't really apply as I am starting trading with 1 contract - cannot go smaller or larger than this with my initial risk capital. And I will need to double my trading size to 2 contracts at some stage, without having out of proportion losses at the new size. After this, compounding should be lower risk due to being able to add smaller increases in size (2 to 3 contracts, 3 to 4 etc).

I am always paying the spread on my simulation account and these results include the commissions and fees I will pay when live.

I don't plan on going live until I know with high probability I can compound without ever drawing down more than 50% of total account equity.

How do traders here approach statistics and position sizing?

 

[CompoundTrader]

Sample size.
The above doesn't tell you how many trades I took. I do 6-10 per day, so am I right in thinking that 20 trading days would be a sufficient sample size (120-200 trades) for a short term day trading strategy?

Here is the stats produced directly by my order entry software for this week. I made some improvements last weekend, so profit factor for this week above 2. I know that I need at least 4 weeks of good stats without making changes to the method in order to go live - but can I be more scientific?

 

[OrderFlowDashPro]

Unfortunately, your question is not easily answerable. More over, unfortunately all of the the statistical tools such as backtest, sample size, walk forward analysis, and historical win/loss ratios have limitations when applied to markets that can make their results non representative of the results you are likely to actually achieve.

It is therefore more appropriate to think of developing trading systems as a craft rather then a science. Skilled system developers use statistical measures as a tool but understand their limitations and most importantly know the results may not carry forward into the future.

As such, the best advice I can give you is to think about the models you create, the assumptions implicit to those models, and how those models may fail in the future. Monitoring the results of the system in real-time and analyzing those results in context to the historical results is the most practical way for evaluating your system.

 

[CompoundTrader]

- I'd advise using leverage of less than 10:1.
Why? If I used this strategy on the COMEX 100oz gold contract, the notional value is some $126,400. This suggests that I need $12,640 to trade a single contract. If my average loss, including commissions, is $100, and more than half my trades are profitable, the chances of drawing down $2,000 per contract is very low. If this happened, I wouldn't want to invest any more in the strategy. Where did those appalling results come from would be my question. I still need margin to maintain the account and take positions down to the max drawdown, so $4,000 a contract is sensible. Why would I invest the other $8,640 when I have no wish to lose that much?

- You are correct to ignore Kelly Criterion for bet size, it doesn't apply.
Ok...but surely something does apply, and that is what I am trying to find. Any ideas? I'm currently reading about calmar, sharpe, sortino ratios but it all seems very abstract so far.

- The data is non-stationary. So in my view 20 trading days is not sufficient. Not even close.
What does non-stationary mean? What would be sufficient? Should I be counting days or trades? If you made 5 ticks per contract traded net of costs, over a sample size of 150 trades, were profitable every day for a month, and your max drawdown was less than 1/5th of your net profit, would you want to test further, or would you risk $2k token trading 1 contract?

Assuming the method consistently identifies a repeating condition, how much data is required to prove that the condition does indeed repeat? Maybe it won't repeat forever, so we need to know when it stops working, I agree. But we also need to know when it is worthwhile to exploit.

- Did you backtest it?
No.

- 50% drawdown is higher than a lot of people could handle.
Perhaps...if they haven't done the maths. This is what I'm trying to understand here. What level of drawdown tolerance produces optimum returns. Total return is the parameter I am trying to maximise.

So lets say I had gone live Tuesday. 50% of the initial $4k is $2k. I am up $1980 so far. I draw down $2k. I lose $20 and find out my system doesn't work. Is that a lot of risk?

Lets say I double the account twice, and then halve it. Is that a lot of risk? I would have only $4k profit. Assuming I value my time at zero for the purposes of this exercise.

Lets say I lose 50% out the gate. $2k. Something is different in live trading, despite all my planning. I've been as thorough as I can be, and the unforeseen or unforeseeable occurred. Is that a lot of risk?

Lets say I trade the account for six months, double the account five times, and then halve it in a single day. End balance is somewhere around $64k. The guy who took the same trades as me at 10:1 leverage or less funds at $12k instead of $4k to trade a 1 lot, and is up only $16k after he has his bad day. So his net profit is $16k against my $60k, his account is at $28k vs my $64k, and his return is 230% against my 1600%. Yet he thinks he is safer because he doesn't handle high drawdowns and trades at lower leverage.

Lets further say that someone has 1 years living expenses and $10k risk capital to make a go of full time trading. Do they want to be allocating $10k to one strategy, trading it with low leverage and low returns, and earning $16k in six months which just covers their living expenses so not enough to reinvest profit? Or do they want to allocate only $2k to a strategy, therefore ability to allocate to multiple strategies or products, and the chance that if the results are as good as the theory, that they are compounding up well after six months so that they are not making big (%age) withdrawals from trading account to pay living expenses?

In what world do I want to be the other guy? Even if he is profitable, he doesn't have enough capital to make it worthwhile trading such low leverage. The only time it would be better, is where the system is sufficiently unsuccessful that it profits at 10:1 but blows up at 30:1. Remember we are talking about short term intraday trading where the max the system will trade is between 30 and 100 lots depending on product and trade type. Eventually, even if successful, leverage reduces automatically as strategy becomes capacity constrained, but you'd still switch it off if it managed to draw down $2,000 a contract no matter how long it had been successful for or how much it made.

For day trading your own account with limited capital your method has to be one mother of a winner. I'm not there yet, and I'd suggest that profit factor 2 isn't high enough for my purposes. Is that the only or best metric to track?

The whole point of this thread is to find the correct threshold of certain stats - profit factor - gross losses - etc where it is sensible to employ this leverage.

How many here are trading for a living, and assuming they did not start with high five or low six figure capital, how did they compound up their accounts? What strategy types and leverage levels used?

Thanks for the replies so far.

 

[CompoundTrader]

Actually, that point above is massive. The guy who takes $12k and turns it into $28k in a year is likely back at a desk job by year two, despite knocking it out of the park in terms of both net ticks and net % gain. Whereas if he had worked to understand and improve his strategy to justify the use of more leverage - a bigger initial risk - he has more chance to succeed and enter a category where he can afford to take less risk.

If he keeps his living expenses small, and gets that $12k closer to $100k at the end of year 1, he is self sustaining and has a shot at a million in net profits by year 5. If your business plan can't support a realistic chance at a million in profits in 5 years, you probably shouldn't be in trading as the effort and work required would earn a better return elsewhere. Just my thoughts.

 

[gerryg]

actually it is difficult to make trading plans (no strategy) and implement them. Forex volatility doesn't put up with schedules or plots and respect constant experiments and investigations. Just keep it in mind and you will success for sure.

 

[MajorMagnuM]

- You are correct to ignore Kelly Criterion for bet size, it doesn't apply.
Ok...but surely something does apply, and that is what I am trying to find. Any ideas? I'm currently reading about calmar, sharpe, sortino ratios but it all seems very abstract so far.
some interesting studies done on dilutions of kelly criterion

 

[forexjudge]

You can go live now. I don't think you should waste too much time with your demo account. If you have a few bucks to spare, then open a micro account and do some real stuff.

 

[robertofx]

You can go live now. I don't think you should waste too much time with your demo account. If you have a few bucks to spare, then open a micro account and do some real stuff.

Micro account is very ideal to test live trading. At the similar time you can learn psychological aspects of trading

 

[gerryg]

Micro account is very ideal to test live trading. At the similar time you can learn psychological aspects of trading

Every trader should start with micro account, because it is essential start for any newbie. I started with micro hotforex acccount with leverage 1:400 and 500 deposit. Quit enough for a start.

 

[Giovan]

no backtests required unless you are using an auto system. The only sure way is to increase your sample size. 6 months to an year testing a couple of strategies at the same time.
pick the most consistent after the test period.

 

[bigmistake]

Go when you can beat demo system easily since you don't have any tool to improve your skill. Real is the best way but it can make you become a loser so be carefully.

 

[NVP]

ive been trading part time since the 1980's and got more serious from the early 2000's with forex

due to impending redundancy I am seriously considering going pro later this year and am testing a few of my favourites at present for daytrading forex

I trade microlots live so am already using real money .......but chumpchange

its pretty simple when you step it up to the big league

1) you have complete and utter knowledge and faith in your trading strategy

2) you have complete and utter control of your trade execution

3) you have complete faith in your broker and platform


thats it ...........
N

 

[NVP]

- You are correct to ignore Kelly Criterion for bet size, it doesn't apply.
Ok...but surely something does apply, and that is what I am trying to find. Any ideas? I'm currently reading about calmar, sharpe, sortino ratios but it all seems very abstract so far.[/I]
some interesting studies done on dilutions of kelly criterion

I find that buying things when they go up and selling when they go down works best in the long term

seriously ....i'm not kidding........ its that simple .....the rest is detail ..thats why marketeeers dont want to tell you that .........how would they charge $000's for saying that ?
N

 

[CompoundTrader]

Update

I've been tweaking my method to take account of the following:-

• fees
• volatility / stops
• risk management

Results attached. Nearly happy with this and plan on going live soon.
Number of contracts traded: 88
Net profit (including simulated commission): $9,143.64 on $25,000 risk capital
Profit factor (gross profit / gross loss): 3.26
Max drawdown: $1,193.20
Total profit to max drawdown ratio: 7.66
Expectancy per contract traded: $103.91 including costs (around 11 gross ticks, this includes losses, costs are typical retail costs around $5/contract including execution platform costs)
(I don't track strike rate as a key metric, but for those interested its 66%)


Fees: my prior strategy was earning 1-3 ticks per contract traded. While this is enough to be profitable, I didn't feel it was sufficient given that I'd be paying away half a tick in fees. There are ways to reduce costs, but only economic for larger / more active accounts. My latest stats show that I am netting 10 ticks per contract traded, including fees and losses. This is much more successful.

Volatility: my strategy would start to lose when market volatility kicked up. I am fully discretionary, and my worst loss day was when I expected 5 markets to go up, bought all 5 at various points in the day, and took losses. What is the reason for this? Stop losses. Calculating a stop size for every product based on volatility, tolerable dollar risk, and an understanding of how "tight" my entries typically were. Of course when I am stopped out by what I recognise is market volatility rather than my position being wrong, I immediately get back in. This opens up the possibility of multiple back to back losses caused by automatic stop even when I'm entering the market at very close to the correct point.

When volatility increased, my usually "safe" stops were being hit by normal market action. This is obviously not sensible. But it led me to think about why I use stops at all, since if the market does not do what I expect after entering trade, and I consider it is no longer worthwhile to hold, I can get flat in a single click.

Are stops really more to do with fear; perhaps even attempts at controlled gambling, than as a tool belonging to a serious and profitable trading business?
Hypothesis: a profitable trading strategy can be rendered ineffective by incorrect stop placement


Risk management: most are told by third party educational vendors to use stop to manage risk. Yet most lose. Add in trading with fear due to being undercapitalised/overleveraged or not understanding the market well enough, and it is simple to see why most do not succeed. Stops are not and cannot be a substitute for risk management.

My risk management now consists of the following:
-understanding the market and my strategy (this is #1 - without this, its attempting to add risk management to something which is, in polite terms, guessing)
-larger account size (I had previously considered that $4,000 was enough to trade 1 lot futures - now I think $25,000 is the minimum for the products I trade)
-therefore not overleveraged/undercapitalised
-keeping risk constant per trade (the contracts I trade are $5, $10 or $12.50 per tick, the volatility varies, and I want to risk the same % of account per trade)
-emergency stops to close out positions in case of e.g. flash crash but they should not be hit in normal trading and not relied on to take me out of position should it be wrong
-maximum position size per product and overall to limit max exposure

I want to write more about keeping risk constant per trade. I was wanting to risk, having small account, only $2-300 per trade. However, some markets I can enter and know if I am correct within 10-15 ticks, and other times market volatility is higher. I knew something was wrong when:

-some profits would be $800, others $200
-some losses would be $100, others $400
-some trades which would make $1200 if I risked $6-800 couldn't be taken at all due account size

So I worked out how much it would be to equalise risk. Risking 20 ticks on 4 lots = risking 80 ticks on 1 lot. With $25,000 equity, I can aim to risk around $800 per trade, and adjust the contract size accordingly. This means I get the same risk profile for each trade whether trading slow calm markets (euro/usd) or fast markets (gold, natural gas).

This also means I can reduce position size when markets are more volatile. Most will blow up doing this if they do not have a clear model for the expected future price move. This only works if the profits taken with the larger risk are worthwhile. If I'm risking 50 ticks I want at least 50 ticks profit when the trade works out.

I'm also thinking in terms of "units of risk". I risk as close to $800 per trade as I can. $800 = 1R. I am expecting minimum profit of 1R per trade. I find that I'll gross around 1R per day traded. This means in practice I can earn around $1k for every $25k deployed per trading day. This number will adjust with volatility, but the key is that when volatility goes down, lot size will increase to have the same dollar risk per trade.

A small account, say $5k, with a limit of 1 lot and trying not to risk more than 250-300 per trade does not have this flexibility, which makes it challenging to trade multiple products.


Any feedback? How are others managing risk? Do you agree with my observations about stops? If you don't use stops to manage risk, then how do you manage it? There is little written for retail traders about this, and its obvious that the "use stops", "risk 2%", etc aren't optimum or even useful.



(and for those wondering, 1 May I made a mistake with the execution of my strategy. Have done some work on it, not likely to happen again. This was simply human error.)