When should I dispose my algo EA parameters
Downtrend started within that initial backtest or out-of-sample?
Give some details on your optimization approach (walk-forward, static window, etc).
What's your key quality metrics (profit factor, max drawdown, etc).
basically downtrend started when i tried it on demo, also we can see that it started at the end days of back testing, before its tested for 4 months it was super profitable even with low dd.
btw my optimization approach is static i guess, with no forward test.
my key quality metrics are actually lowest dd possible.
4 months is too short to get robust output.
I'd encourage extending your backtesting for a longer period (4-5 years to cover several 'life-cycles').
Be prepared to constantly re-optimize.
NOTE: Don't expect anything consistently profitable. You get what you pay for.
IMPORTANT: Don't get discouraged and never give up!
maybe you can use the previous parameters with 0.01 lot size on the first 1 - 2 days of a new month or after big news/economic data release. so if you sacrifice 1 - 2 days to make small losses and small gains, then you know how to optimize it (whether it needs to be optimized)
that sounds like a pretty good advice, momentum usually shifts every month or after big news release as my strategy is based on 1 m tf, it really needs to be constantly changing or changing parameters.
thing is when i optimized it.it gave me lots of parameters and i selected the top one whos most profitable and when i put that on demo this whole month, it turned out to be losing, and when i back tested the less profitable parameters, those turned out to be profitable this month.
kind of i need to consider more factors too, maybe in the future applying it to neural network will do the job more efficiently i guess. btw thanks.. :)
4 months is too short to get robust output.
I'd encourage extending your backtesting for a longer period (4-5 years to cover several 'life-cycles').
Be prepared to constantly re-optimize.
NOTE: Don't expect anything consistently profitable. You get what you pay for.
IMPORTANT: Don't get discouraged and never give up!
i heard that minimum 1000 trades back tested is enough for maybe a couple of month as it is a 1m timeframe strategy means 200 average trades a month, 50 a week. i actually back tested it on previous year and it turned out to be losing in 2023 but this year it was on fire.
it will take ages to optimize it for 4,5 years but i should do it. will build a double intel xeon 22 cores system for specially optimizing strategies. btw how long do u have to optimize ur strategies for algo? also thanks :)
If algo trading was easy, everyone would be doing it profitably.
There are no shortcuts - keep digging in until you're happy with results. It's a long-haul.
The process requires patience and perseverance. Some day down the road (likely not soon) the sun will start shining on your algo street.
Happy trading!
🙂
If algo trading was easy, everyone would be doing it profitably.
There are no shortcuts - keep digging in until you're happy with results. It's a long-haul.
The process requires patience and perseverance. Some day down the road (likely not soon) the sun will start shining on your algo street.
Happy trading!
🙂
obviously its not easy, but i feel like theres a light in the dark cause im trading from 8 years manually.
99.9% traders i have seen are manual traders. theres not even much info online about algo, i found it interesting tbh !
the other 0.1% are mostly bigger firms who own super computers for bactesting and optimzations.
the major problem i have seen with manual traders is mindset and patience which is already eliminated here i guess but yeah its gonna be a long haul cause theres too much data i have and i dont really know how to utilize it effeciently also no coding skills too :( but that doesnt matter that much.
happy trading to u too :)
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
hi there mlq community,
i m new to algo trading, recently found some open source bots, tweaked it, optimized it and they are giving pretty good results on real tick data, my strategy/bot is based on 2 moving averages crossovers
and i optimized it for 1 minute timeframe on gold, back tested at least 4 months of data approximately with a 1000 trades record and results are pretty good, as markets are constantly changing, i heard
that we need to dispose off the previous optimized parameters and reoptimize it further.
Point here is WHEN?
a lot of u must be familiar with with this and please of someone can guide me?
here are the results of back testing.
we can see here it performed well for 4 months of 2024 but it went into down trend from the start of 5th month