When do you accept a strategy as profitable and viable?

F

Fooze666

I'm new to trading and have been working on a few strategies trading paper for a few months now.

I have one strategy that I believe is going to be profitable and another that is looking good but still too small of a sample to say anything certain.

I'm using ThinkorSwim and have had some difficulty programming strategies for backtesting. The way the scripting language works doesn't seem to make it possible for me to program what I want to do exactly (such as buying in the middle of a bar) and results get very skewed. So, I've developed a preference for testing through paper trading.

I want to move into real-money trading on a fairly small level to see how it goes with at least one of these strategies.

My question to you guys is...When do you accept that a strategy you're developing is solid enough to become part of your regular toolbox? For example, do you go for a certain amount of profit over time? Or maybe you want to see a minimum of 100 trades and gauge profit and win rates?

I'm not sure where I can stop and say, "Ok, this looks like it's ready to give it a try with real money on the line."

Also, as a secondary question: What do you limit your risk to on each individual trade? I was planning to trade with a limit of no more than 1.5% of my funds at risk on any single trade. I've seen a few posts here and there though saying maybe that is too much.

Thanks in advance for any opinions you can offer.
 
There is a lot more to do. Think of it as you just built a race car, your first. Now it is test drive time. Then a mock race or 20, then your first race. You need to integrate your execution tactics into the strategy. Then plan for all the conditions the market will throw at you besides normal variations. Then there is execution issues such as hardware and software. Make sure those have contingency plans and you can implement them easily and quickly.

Specifically, trade on Sim, then trade smallest possible amount that does not change the tactics. Then go back and have a HUGE set of evaluation benchmarks, which you already defined, to measure the strategy. Then revise and tune. This applies to your evaluation benchmarks too. Repeat until it becomes VERY predictable. Hope that helps.

PS: as a data point I have been developing an ES strategy on SIM. Been live 6 times or so over 18 months. Going into the March contract now (on SIM) and it looks like this will be my final test run. Test runs are 3 months. Just a data point on ES which is brutal on a retail platform and internet connection. (Also have 6-7 years prior experience in futures live trading, a long time ago, but the market and tools have changed, so only somewhat relevant)
 
Last edited:
do you use limit orders? are you entering and exiting on the same bar?

do your backtest include commissions, lost interest rate and some slippage?

are any of your trades placed on the close of the session?

what is the drawdown and net win/loss as well as the percent profitability.

does it trade both side of the market or fitted for a bull market only?

how large is your sample size the test need to have many thousand of trades to be valid otherwise you may have found a random pocket of curve fitting. 20 trades for example is not acceptable at all, 2000 trades is getting there.
 
Animated Gif analogy:

What I thought my strategy would be. Orca.gif

What it ended up as B.gif
 
@Fooze666, this is a very good question.
It's very personal. Each individual has its own goals and expectation.
For me, I have a minimum net profit per week per contract (I trade the ES) as a goal before I'm using that new strategy live with real money. That minimum net profit must be obtain at least 4 weeks in a row in sim.

Your second question is also very personal. It mostly depends of your risk tolerance.
For me, my risk limit is set at 1% on each trade.
 
I'm almost exclusively long-term trading forex off daily charts. When considering a new or revised strategy I tend to look back for a year (but excluding 2020). The strategy must have a positive win rate or I probably won't do any further work on it. Yes, I know negative win rates are viable if you have a high r:r - have to draw a line somewhere.

When using the strategy I want the majority of months from the last 6 to be winners with no mega-loser in the mix.
 
Backtest, carry out simulations using monte carlo, boostrapping. Deploy it in paper account for a period of time and monitor performance. If performance looks fine, proceed to deploy and continue to monitor for any signs of strategy decay (you can use Kolmogorov-Smirnov Goodness-of-Fit Test for this)
 
You must log in or register to reply here.
Back
Top