Hello all,
I’m running some simulations, to test software I’m writing and I’m unsure whether my results are expected or not. It would be great to get some advice.
Setup:
100 traders ($100k start fund), each opening a trade 1 every 10 minutes, on any of the major pairs, all opening trades at random.
Each trade is either long or short, and opens with a TP/SL set at 0.3% of the current price.
Trades run until they hit TP/SL
Lot size is 0.5 per trade.
Leverage is 50*
Over time I am seeing nearly all traders are losing money.
Currently it;'s been running about 9 days, and total loss is -460911
Is this expected because of the spread, or is this way off expectations?
Its a worthwhile experiment, it can throw up some useful statistics, it can help in money management discipline.
But as the entries are random, the SL’s a re random and the TP’s are random, I guess the current performance is random also, though the depth of the drawdown is due to your selection of leverage.
The SL and tp are always a consistent percentage of the trade price.
I’ve run it multiple times and there is a consistent downward trend in all cases.
It may be a bug in my software, but I’d imagine there would inevitably be a downward trend due to the spread but I don’t know if this rate of lose is what that would be responsible for…
Yes, spread would always cause a negative weighting to your results. Its possible that a narrow 0.3% SL/TP would be inside the spread on some occasions on some markets, so perhaps the trade would close immediately at the SL if that occurred?
Your loss stats is worrying because it doesn’t add up to what could and should be be a much closer random result. I would expect a P/L % return to be between c.53% - 47%.
The spread would certainly affect the win/loss ratio, but not by much. A better SL/TP could be 0.3% - 0.45% of price or even give it more S/L breathing space - say between 20 - 30 pips, which could be why you’re constantly losing.
Also, record buy trades against sell trades to see if there is any noticeable difference between buying and selling results.
I would also split the trades by days of the week to see which days could be more profitable. And which pairs are more profitable - which is important when not randomising…
And also see what time frame works better.
All in all, though, a very worthwhile experimental exercise. Well done. Keep posting your results which I find very interesting.
The first is I think the TP and SL should be set equally above and below the Bid for a buy (and Ask for a sell) so that there is an equal distance the trade needs to move in either direction to close. This should give equal probability of numbers of winning/losing trades. (currently I’m using the price)
BUT the size of the losing trades should, on average be a little bigger than the winning trades.
Back to example:
Winning trade = TP-Ask = 0.00215263
Losing trade = SL-Ask = 0.00219263
Which equates to a return of about 98.2% (I think)
I’m going to run this sim with the changes above and see what happens… I’ll also increase the 'breathing space for the TP/SL as you suggest - this shouldn’t affect the percentage of wins:loses or return rate.