LACMA performance thesis

The Quest for the Ultimate Trading Signal – and how it birthed the FXDashboard AI risk engine

The core insight: LACMA was a machine learning consensus algorithm, engineered to parse real-time trading data from thousands of active funded traders. By weighting each trader's implied prowess in 4 classified market conditions, and generating entry signals for each market condition, it generated consistent returns with near-zero risk of ruin—all on just 2x leverage. For two years, LACMA delivered extraordinary 3-5% monthly returns with zero losing months (before fees). This peak performance was achieved exclusively when the model was trained on real funded subscription accounts (2017–2019)—traders who paid monthly and traded with genuine seriousness. Once the data source shifted to post-stage-1 challenge demo accounts, signal quality collapsed. Losses appeared. The company pivoted. Today, the same AI that once sought the ultimate signal now provides traders with human like mentoring and signal provision, and at the same time protects prop firms as the FXDashboard risk engine—trained to detect and block the toxic behavior that poisoned its own data.

Subscription-era (real money)
Monthly return+3% to +5%
Losing months0
Avg profit factor2.3
Data sourcereal funded
Demo-challenge era (post-stage1)
Monthly return-0.2% to +0.7%
Losing months6 (of 17)
Avg profit factor1.12
Data sourcedemo accounts
17 months of actual LACMA performance (Jan 2022 – Jul 2023)
2022 real-data months Jan–Aug (except Sep/Oct missing) – strong performance
?? 2023 demo-affected months erratic, negative prints
H1 2022 (peak real data)
  • Apr: +2.92% (119 trades)
  • May: +3.29% (236 trades)
  • Jun: +1.11% (408 trades)
  • Jul: +2.58% (199 trades)
  • Profit factor often >2.0
Late 2022 – transition
  • Aug: -0.36% (first loss)
  • Nov: +0.71% (65 trades)
  • Dec: +0.36% (40 trades)
  • Signal starts to weaken
2023 demo-dataset collapse
  • Mar: -0.01%
  • May: -0.26%
  • Jun: -0.39%
  • Jul: -0.15% (only 7 trades!)
  • Profit factor drops below 1
Winner % over time

High win rate (>60%) during real-data phase; drops below 50% in mid 2023.

Avg trade size (lots)

Real accounts traded larger size, demo accounts became erratic.

Net pips monthly
Valley (max drawdown pips)

Valley deepened in 2023 as toxic flow increased.


Why this matters for FXDashboard

LACMA's neural network learned optimal behavior from real funded traders who paid a monthly subscription – they traded seriously, without gambling. When the industry shifted to demo challenges, the data became polluted with "toxic flow": martingale attempts, gaming, and unrealistic risk. The signal collapsed.

Liquid's insight: Instead of giving up on AI, we repurposed LACMA to detect and block that exact toxic behavior in real time. The FXDashboard risk engine is the direct descendant of this painful lesson – it identifies the "spirit" of rule breaking, even when traders try to game the system mathematically.

Today, FXDashboard protects prop firms using the same AI that once generated 3-5% monthly – now applied to risk management. And provides support, mentoring and personalised insights as well as trading signals to individual traders. It delivers non halucinatory and fast analysis to managers and traders using Liquids own trading view platform or MT5, MT4 and any other platform with an API. It will soon allow any trader at any broker to connect to our AI to receive the benefits of personalised advisor trained on 5 years real account data. Lacma will now do all risk management, onboarding, KYC checks, violations, emailing and trader management on a pay as you go basis for a new broker or prop firm owner or fund manager. lacma is connected to all the markets as well as social media and can deliver news, facts, gossip and sentiment from the markets

Underlying monthly data (provided by Liquid Fintech):

Month Net% Profit factor Win% Trades Net pips Jan22-0.210.9861%38-702.8 Feb22+0.521.1066%58452.4 Mar22-0.051.0752%82-198.8 Apr22+2.922.2960%1193267.1 May22+3.293.7169%2365068.8 Jun22+1.111.4055%4081270.4 Jul22+2.582.0359%1992206.9 Aug22-0.360.7654%134-1702.9 Nov22+0.718.6395%651470.5 Dec22+0.362.6568%40365.4 Jan23+0.141.3071%31247.1 Feb23+0.061.1268%2276.7 Mar23-0.010.9971%631360.8 Apr23+0.582.9864%55606.9 May23-0.260.7748%112-337.7 Jun23-0.390.5945%118-979.3 Jul23-0.150.1343%7-244.2
* Data series includes only months where records exist. The drop in mid-2023 directly correlates with reliance on demo-challenge datasets.