This is the best I can make of EURUSD: image At least equity is going North, but certainly not as convincing as the averaging down equity curves we have seen.

The Myth of Averaging

Level 3 optimization gave: L3_ScaleInPoints=200, L3_TrailingStopLossPoints=375. image I am giving up... Since AUDCAD is fairly ranging, perhaps this is not the best symbol to add to winners and follow trends? Let's try EURUSD now...

The Myth of Averaging

Thanks. Good to hear you are able to add into winners. So far my experiments with averaging up are not hopeful as you can see.

The Myth of Averaging

Next step is the Level 2 settings: image I am a bit disappointed in what this gives me: image image Hardly an improvement over Level 1... The "best" setting is L2_ScaleInPoints=800, L2_TrailingStopLossPoints=375: image image

The Myth of Averaging

Now we are going to hunt for a good L1_TrailingStopLossPoints: image This gives the following results: image image Hmm, this is awkward. The best is L1_TrailingStopLossPoints=10, but we don't want to build a scalper EA that is trailing at 1 ...

The Myth of Averaging

Ok, let's start trying out this thing... First let's choose a good time of day to enter the market. We optimize "Balance Max" for EntryHour: image All terrible performance and not much difference: image So let's choose EntryHour=10 for now and ...

The Myth of Averaging

Averaging UP Ok, new chapter in this exciting journey to prove the mathematicians wrong... I have uploaded a new EA in post #1 that implements Averaging UP (adding to winning positions) instead of Averaging Down (adding to losing positions): ...

The Myth of Averaging

Thanks a lot for trying to reproduce the results! Great to hear AUDCAD gives you the same results!! So this thing actually works, and is not a fluke... Amazing For EURUSD I used different settings. Perhaps you can try yourself what settings you can ...

The Myth of Averaging

To verify this was not just a fluke, I tried again using a symbol that is less ranging and typically not used for grid trading: EURUSD. Guess what: image image This confirms we can really push the profit distribution profile to the left and get ...

The Myth of Averaging

You confirm what I always believed and what is taught by the mathematicians. And yet, the backtests suggest that during a range of many years the random strategy is profitable without any Black Swans?! Absolutely. I don't recommend actually trading ...

The Myth of Averaging

Indeed interesting to read his story and approach. Said he made a lot of money using averaging down (by hedging using another pair) and was trading full-time. Has not been on FF the last 5 years. Hope he did well and is retired by now. I am fully ...

The Myth of Averaging

Now that we investigated Averaging Down, perhaps it is interesting to investigate Averaging Up (i.e. adding to winning positions, rather than scaling-in when in drawdown)? What do you think?

The Myth of Averaging

Yes, that is the intended behavior. Indeed, with these settings a Level 4 basket of trades is apparently closed in loss. As long as the other levels (that we get more often) make up for that we still get a rising equity curve.

The Myth of Averaging

Yes, I agree! As a scientist I always blindly believed the mathematicians. I expected to prove to ForexFactory that you can not push the profit distribution past the zero-point. I was wrong. This result is totally unexpected!! The problem with more ...

The Myth of Averaging

Now that I found a way to backtest 100x faster, I decided to not only add a Level 4, but also revisit the previously found settings as these all somewhat affect each other. Since we now have a rising equity line, I switched from optimizing "Balance ...

The Myth of Averaging

WOW... I just had a great idea! What if instead of backtesting each setting with 100 different random sequences and then calculating the median profit, we apply the new version of the EA that runs ALL random strategies simultaneously? That would be ...

The Myth of Averaging

If we would run the EA live with the settings found in post #25, what profit do we get? The graph above is a histogram showing the distribution of the profit of 100 backtests each with a different random buy/sell sequence. While -50 is the most ...

The Myth of Averaging

The search for the Level 2 SL/TP is finally complete after almost 2 days. That took a painfully long time :-( Recap: I let the MT5 optimizer not only try several SL/TP combinations, but let it try these combinations with 100 random buy/sell ...

The Myth of Averaging

No, not yet. That would almost start looking like a grid! ;-) BTW, backtest for L2 is still running. I'll keep you posted...

The Myth of Averaging

Yes, it is always nice to see that input parameters are not too sensitive! In general, select input parameters in the middle of a "plateau", not on a sharp "peak"; even if that means a bit less profit. Your strategy will be more robust for future ...

The Myth of Averaging