Whats your favorit method for EA regime detection???

Good morning everyone!

I wish you all a happy new year, lots of profits, good health, and plenty of motivation!

I'm starting this thread with the question:

What are your best methods for distinguishing trending markets from range-bound markets?
The question is about automated trading with Expert Advisors (EAs)!

Give me all your technical suggestions.

The best answer will receive an EA from me that has generated almost 100% profit on EURUSD from November to today with a relatively low drawdown on the M1 timeframe.

The EA has really good money management features, but there's still plenty of room for improvement (compounding, recompounding, regime detection)!

I know the sample size is relatively small, but whatever!


I'd like to rephrase the question more, but my head is still hurting from last night

Best regards,

str8v4lu3

Okay, one possibility would be to use a TMA with an SLWMA or another moving average. I got this idea from another very informative EA thread, and I think it's truly an outstanding idea. The problem is that this TMA indicator, which always looks so nice, repaints like an artist. I haven't tried it in StrategyTester yet. But I think it's worth a shot.

If SLWMA is < TMA lower Band = Trending long
If SLWMA is > TMA upper Band = Trending short
If SLWMA is between upper and lower band = Ranging Market

I've had bad experiences with combinations of ADX and other indicators so far. If the EA trades on the M1 chart, perhaps an MTF ADX indicator combination would be a solution.

As I said, I don't want to answer my own question here, so I'll wait for your ideas...

One method I read about is called Marcov Regime Detection. I tried it once but didn't pursue it further. Definitely worth a try though!

Marcov Regime Detection

I have no idea if it's being used correctly.

no one interested in this thread?

Happy New Year! For EA regime detection, I usually combine a moving average envelope with an MTF ADX. When price is consistently above or below the bands and ADX confirms strength, it signals a trending market. When price stays within the bands and ADX is low, it’s range-bound. I’ve found this simple approach works well on M1 and higher timeframes.

Happy new year too my friend : )

Thanks for your help, that's great.

That's a really helpful answer.

I'll test it.

I think MA Envelopes MTF would be a good fit.

But I need several days to test and integrate

You're already a candidate for the EA : )

No EA required but the solution used by most quants is gnarly.

In FX, HMMs don't work well. Unstable, no time of day memory, work worse than rolling vol filter. So I'd park that. And the answer?

The aim is to ID - normal vs unstable, trend vs reversion. In short, you need to predict what's expected, monitor against what's happening, then how far apart these two things are, add some persistence, then flip state.

Trend and reversion are the easy ones: create an efficiency ratio over n steps. Normal and unstable are way trickier - you need hourly vols, an hourly consumption curve (because vol is not a straight line in an hour), compute realised variance (ie create a speedometer), and then add persistence so it doesn't flick on and off like a light bulb. You're basically saying unstable markets are either running cold (way less RV than there 'should' be for the time of day and pair), or hot (too much RV compared to forecasts). Add in the trend state, and then gate your signals accordingly

The difficulty isn't the maths here, you can AI all that, but in generating decent hourly vol forecasts which need to include high impact events - because we all know what NFPs do to calm markets.

The pay-off for all this though is 100% worth it. Most strategies improve with decent regime ID and it's *the* bedrock of most models in some form or other.
Added bonuses; the hourly vols have their own use for sizing and probability zones of reversals/trend if you calculate whether vol consumption is speed up or slowing.

I know this only because I'm knocking up the same in between trades. About 1/3rd of the way there so far...