Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
I've been working for a few months on a strategy for NT8, the strategy I came up with runs on the /ES 15M chart and each position is 1 contract. I'm looking for your input in terms of analyzed information, am I on the right path? would you consider this a decent % that's worth running in live environment? what statistics should I strive for?
Thanks for your time
Can you help answer these questions from other members on NexusFi?
Broker: NT Brokerage, Kinetick, IQFeed, Interactive Brokers
Trading: ES
Posts: 159 since Dec 2014
Thanks Given: 40
Thanks Received: 166
I'd recommend to rerun you Monte Carlo and remove the top 10% winners but leave all of your losers. This will ensure that you don't have a few mega trades accounting for the majority of your profit. I also always try to look at ticks instead of $$$ summary but that's more personal preference. As your screen name suggests, if this is crude oil, are you getting unrealistic results on the Wednesday inventory releases? I've had that throw a few back tests before on CL so beware.
That sharp ratio would concern me which is why I recommended the MC tweaks above.
Curious, subscribed
edit, completely missed the subject line saying this was ES, disregard CL comments.
My suggestion would be to run it with real time data in simulation mode during the day. Then run a back test on the same time period to make sure the numbers match. Its been my experience with NT that they give you better fills when back testing vs actual trading.
Thanks for the suggestion. Once I remove the top 10% winners, it turns into a big loser. However, my strategy there is based on cutting losses fairly quickly and letting the profitable trades run a bit longer, so I'm not sure if this setting is actually a fit for this strategy, but interesting nevertheless. I have a few revisions of my strategy, some with a smaller sharp ratio, I'll do further testing tomorrow.
Sorry for the links, I'm unable to post images with less than 5 posts for some reason.
i.imgur.com/jDAmP9t.png
i.imgur.com/pbmJZ5J.png
Would that be concerning?
Thanks Robert, I have been running it 24/7 (paper) on an Azure instance the past 30 days, so far so good.
For example: Was this your first attempt with this ES strategy, or was this the 502nd iteration working with ES? That makes a huge difference in believe-ability.
You also included $0 for slippage, are you using limit orders to enter AND exit? Why no slippage?
And if you are using limit orders, are you assuming fills on touches?
The important point to notice: I'm not even mentioning the statistics you got, because the actual numbers aren't all that important. Things you did to generate those numbers are far more important IMO.
Trading: The one I'm creating in the present....Index Futures mini/micro, ZF
Posts: 2,311 since Nov 2011
Thanks Given: 7,341
Thanks Received: 4,518
Backtesting is a skill you have to develop. The skill in and of itself is unrelated to trading. This is counter intuitive.
What you have done here is poor backtesting. You need to go back in time more. Testing chunks of data and walking it forward etc etc. I am no expert but have done it. I find discretionary trading beats the pants off automated any day, so I stopped backtesting.
There are people here like @kevinkdog that are extremely good at it and it serves them well in trading.
I read the above mentioned fellow FIO members book on how to backtest to learn. Find it on Amazon.
Ron
...My calamity is My providence, outwardly it is fire and vengeance, but inwardly it is light and mercy...
The steed of this Valley is pain; and if there be no pain this journey will never end.
Buy Low And Sell High (read left to right or right to left....lol)
I looked at an ES chart just now, and the period you are testing for, Jan 2019 through August 2020 was a spectacularly non-typical time in the market. There was a gigantic move up until Feb 2020, then a huge drop in Feb-Mar, and then back on the long gravy train after that, up until the present. If your strategy decided not to be long for one brief period, you would have cleaned up, as it appears you did, on the backtest anyway.
Would you have done so in real life? And more importantly, will you do so, in real life, on the next drop? How does your strategy navigate moves of many different kinds, and can you count on it doing so again? I don't know the details of how it works, of course, but the question is whether it can do so the next time, and then the next time, etc. You really can't say you've tested that component of the strategy (the not being long in a big drop part ), since it has had only one instance during the test period.
Do a whole lot more, and don't pre-select a favorable time.
I don't do backtesting, because I don't do automated systems, but you should be able to say whether it is reasonable that your strategy will withstand many different market conditions (how about a pure ranging market, with no big up or down, etc.?), and this means you should test for them.
It's been mentioned that there are books on the subject. Kevin Davey ( @kevinkdog) has one that many backtesters have mentioned favorably. Not a plug, because I don't do backtesting, but an observation about what others have mentioned over the years.
But to answer your original question, no, I don't think you have tested enough.
Bob.
When one door closes, another opens.
-- Cervantes, Don Quixote
It may be my 200-something iteration of this strategy. I've spent months trying to get it a bit better and I'm not going to stop I've came a long way, my first strategy (ToS) was perfect on paper, until I realized I used FL events, for example Close[-1] (bear in mind it was my first time).
This is a long-only strategy, it knows how not to get into trades when the market is red. I'm also working on a strategy that does the complete opposite and shorts the red markets, but I found that if I add LONG and SHORT to the same strategy, I'm not maximizing the profits I can get from either LONG or SHORT, when the two are separated, I get better results.
I'm only using market orders in the strategy at this point.
I've downloaded market replays of the ES (limited to 6m chunks at a time) and the results are fairly close to the back-testing results.
For example: There was a gigantic move up until Feb 2020, then a huge drop in Feb-Mar
Feb was above avg month, finishing with 5k profits (11.5k gross win vs 6.2k gross loss) averaging 46.15% winning trades.
March was the best month for this strategy provided between 30-40K (vary based on which version of the strategy I'm back-testing with) in back-testing profits. Averaging 64% winning trades.
I took a month off my real day job, in order to work on back-testing skills and trying to bring this to a scenario where I'm comfortable to jeopardizing hard earned money for it.
In your opinion, do you think that a strategy that was profitable at 2017 will still be profitable in 2018, 2019 and 2020? Based on my testing (and of course, I could be wrong, hence I'm seeking other eyes on this) and market conditions obviously, certain trading techniques can still work while others don't, automated trading is a constant struggle, what's working today may not work next year or v.v.