Backtesting of strategy before going live

Hello traders
I’m backtesting a strategy right now before sending it live, it’s giving positive results for now and has done when backtested for 5 years.

I’m wondering what criteria people are using to determine wether a backtested strategy is ready or not.
How many years do you go back?
How many trades minimum do you want to see?
What win/loss ratio do you need?
What maximum drawdown will you allow before saying it’s too volatile?
What minimum annual return as a %age do you want to see to make it worthwhile?

What other criteria would you use to ascertain if it’s worth going live or not?

Thanks for your time and look forward to reading any input or advice you can offer.

I work on a 100 trades

5 years is enough, assuming more than 1 trade a month.

Win/loss ratio is an organic part of the strategy along with r:r, fixed stop-loss, trailing stop-loss, b/e stop-loss, TP, pyramiding etc. Not helpful to build a strategy so that any one of these parameters is great - it needs to be treated as a whole.

I look for a strategy that has the potential to double the account every year. Without the potential to do that, it isn’t worth the risk.

An issue with some backtesting is that if you have a system that for example trades the USD/JPY, you need to backtest realistically. That is, if you get a good entry signal and go long USD/JPY, then before you get an exit signal you get another entry signal then another then another - you must not count that as 4 good trades - unless you really intend to be massively overweight in USD/JPY with 4 simultaneous longs.

Likewise watch out for counting long trades and short trades in succession when in reality the long would have still been running though it had not hit its stop - there’s no profit in being both long and short the same market at the same time.

Thanks for your reply Tommor
Doubleing the account every year!! I’m a little bit off that, currently i’m looking at between 20% an 30% per annum, with minimal drawdown over my 5 year period, but good to know so I can try up my game a little.

I would be counting your example as 4 winning trades, simply because it’s an automated system. Sometimes i’d be in 4 winners at the same time and sometimes 4 losers. Assuming I’m never exposed to more that a predetermined amont of my account would these 4 winners/losers present an issue?

Thanks Lang15

1 Like

If your real-life trading involves adding a new position to every eligible market in which you get a buy signal, you will find yourself massively exposed to unexpected corrections. A single buy on USD/JPY with a 2% risk means if USD drops suddenly you can only lose 2% of your account. But imagine the effect of a Trump tweet if you’re running 4 buys on USD/JPY, plus 3 on USD/CAD plus 2 shorts on EUR/USD plus a short on GBP/USD…

Your backtesting needs to be realistic.

My 100% profit per year is realistic. Its only 6% per month, compounded. The point is in trading its easy to go broke in 1 year, so that’s the ultimate risk. Therefore the potential reward must at least match it - this is a rule we all apply to individual trades, they should all have an r:r of at least 1:1. So if your trading can cost you 100%, it must also have the capability to make you 100%.

Thanks for the advice Tommor, i’m solely on EUR USD for now with a maximum of 2 trades placed per day, I understand what you mean about excess exposure and will consider that more when I put my strategy to work on multiple pairs.

I’ll see if i can tweak some parameters, risk reward etc to get my return a little higher in the meantime.

Thanks again for your help

1 Like