Quick follow-up on our incremental Ready-for-Live analysis
After finetuning and finalising the workflow, we’ve now validated the incremental “Ready for Live” model with a new visual layer (in attachment):
a persistence matrix that shows, month by month, which EA strategies remain in the top tier — and which ones drop out as more history is added.
This chart combines two dimensions:
• Vertical axis: monthly ranking based on our Combined Score
• Horizontal axis: the persistence of each EA across six cumulative snapshots (June → November)Even with >700 strategies in incubation, and roughly 10 “Ready for Live” candidates each month, the pattern is extremely clear:
• Only 2 EAs stay consistently at the top across all months (true Elite)
• 2 additional EAs show strong but slightly more volatile long-term behaviour (Tier-2)
• Many early high-performers collapse once the historical window expands (typical overfitting)
• A few strategies appear only in the last 1–2 months (late bloomers → still under observation)Key takeaway:
persistence across cumulative periods is a far stronger robustness signal than any single-month combined or weighted performance.With this step, the incremental Ready-for-Live workflow for Group 1 is now fully validated. Each month, a new incremental snapshot will be added to maintain both perspectives: monthly performance and time persistence.
If anyone is interested in the methodology (snapshot logic, scoring, or persistence rules), I’m happy to share more details.
Vincenzo
Hi Vincenzo,
Interesting direction, especially the focus on grid behavior and level tracking that’s definitely a relevant angle.
One thing I’m trying to understand better is how you evaluate these results at this stage.
Most of the strategies shown seem to have a relatively low number of trades so far, which makes it a bit difficult to assess consistency or robustness. I assume this is still in an early phase?
From my experience, and also from discussions with other algo builders, performance is usually only considered meaningful after a certain amount of data has been collected often around 40 to 60 trades or more per strategy. Below that, it’s generally still more about observation than confirmation.
Also, since you mentioned these come from original non-grid strategies, it would be useful to see how those baseline versions perform over the same period. That comparison would give more context to what the grid is actually changing.
The grid-level analysis you describe is probably the most important part here. Without visibility on how often deeper levels are reached, it’s hard to fully understand the risk profile behind the current results.
]]>After finetuning and finalising the workflow, we’ve now validated the incremental “Ready for Live” model with a new visual layer (in attachment):
a persistence matrix that shows, month by month, which EA strategies remain in the top tier — and which ones drop out as more history is added.
This chart combines two dimensions:
• Vertical axis: monthly ranking based on our Combined Score
• Horizontal axis: the persistence of each EA across six cumulative snapshots (June → November)
Even with >700 strategies in incubation, and roughly 10 “Ready for Live” candidates each month, the pattern is extremely clear:
• Only 2 EAs stay consistently at the top across all months (true Elite)
• 2 additional EAs show strong but slightly more volatile long-term behaviour (Tier-2)
• Many early high-performers collapse once the historical window expands (typical overfitting)
• A few strategies appear only in the last 1–2 months (late bloomers → still under observation)
Key takeaway:
persistence across cumulative periods is a far stronger robustness signal than any single-month combined or weighted performance.
With this step, the incremental Ready-for-Live workflow for Group 1 is now fully validated. Each month, a new incremental snapshot will be added to maintain both perspectives: monthly performance and time persistence.
If anyone is interested in the methodology (snapshot logic, scoring, or persistence rules), I’m happy to share more details.
Vincenzo
]]>we’ve been running a very large EA incubation setup since January 2024 — around 20 demo accounts, currently hosting 700+ active EAs, all of them 100% EA Studio–generated strategies (no manual coding, no third-party bots, no external licensed EAs).
Recently, we completed a stable workflow that generates a “Ready for Live” list each month based on FxBlue data and a proprietary scoring system. But we also wanted to answer a deeper question:
What happens to each EA Studio strategy when the historical window expands month after month?
To find out, we did something more.
✔️ We rebuilt the Ready-for-Live list backwards
from June → November (6 cumulative snapshots)
✔️ Then we flipped the perspective
Instead of looking at the list for each month…
we tracked each EA strategy across all months.
This allowed us to see:
• who stays consistently strong
• who collapses when more data is added
• who is a “meteorite” (looks good early → fails later)
• who is slowly becoming a strong EA
• who is just lucky in one snapshot
To measure this cleanly and repeatably, we created a new metric:
⸻
Maturity Score v2
A composite indicator based on:
• how many months an EA appears
• whether those months are consecutive
• how stable PF, Win%, Recovery and Score are
• how performance changes as we add more history
• total trade depth (a major signal)
This produced some very interesting results.
⸻
1. Only two EA Studio strategies are truly “elite”
(6 months stable, high trades, strong KPIs, no collapse)
① EA 250083819
The best EA in the entire incubator.
Always Ready, ~470 trades, extremely strong PF/Win/Recovery.
Total structural stability.
② EA 1506928733
Also present in all months, very reliable, very stable.
Clear long-term maturity.
⸻
2. Two EAs are strong but not (yet) elite
③ EA 1987477025
Very good metrics but more volatile.
④ EA 1074696949
Excellent KPIs but started appearing only recently.
Could become elite with more months of stability.
⸻
3. Three EAs are “mid-tier survivors”
Stable but not exceptional → good as satellites, not for core allocation.
• 1264298459
• 1326733007
• 1958683245
⸻
4. Four EAs are “meteorites” — strong early, collapse later
These looked excellent in June–August but fell apart when more history was added:
• 2120452330
• 1784140930
• 1699341130
• 1740201819
This is a classic overfitting pattern.
⸻
5. Three EAs are late bloomers
Too little data to conclude:
• 1223417011
• 297120297
• 1905225350
They need more months.
⸻
What this analysis shows
✔️ Persistence > peak PF
Many high-PF strategies disappear once we expand the window.
✔️ Second-year strength matters
Real robustness only shows up after multiple cumulative periods.
✔️ Trade depth is essential
Strategies with 200–400 trades behave differently from those with 50–70.
✔️ Our EA incubator filters aggressively
Out of 700+ EAs…
only 2 show elite long-term stability.
This is a sign of good filtering and realistic expectations.
⸻
Our new “Live Portfolio” will include
Core
• 250083819
• 1506928733
• 1987477025
Watchlist
• 1074696949
Satellite
• 1264298459
• 1326733007
• 1958683245
Reject
4 strategies that collapse under expanded history
Observe
3 late bloomers
⸻
I strongly believe that in this forum there is a lot of knowledge and many smart people.
The outcome of the above analysis came up thanks to an intensive exchange with people met in the forum.
A big thank to Hez for the great contribution and inspiration.
This should really encourage everyone to share knowledge, challenges and pain points as well as great ideas.
I’ll be happy to tell your more about my Incubation Strategy...
Vincenzo
]]>