Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

Trading Performance Analysis: GATE SPX500

(we still have one more day of trading)

https://www.forexfactory.com/bl4ckp3n9u1n/264212

My goal is to bring the account back to break even next week as this started off with $200. At the time of posting the Balance is $187.57 (-$12.43 profit)

1. Trading Activity: The system executed a high volume of 71 trades over five days. Lot sizes were small, typically ranging from 0.01 to 0.03, indicating a scalping or high-frequency strategy.
2. Strategy Behavior: The trading log reveals several distinct exit reasons:
   1. [tp ...]: Standard take-profit orders, which accounted for most winning trades.
   2. [sl ...]: Stop-loss orders, responsible for the two largest losses of the week (-7.95 and −7.95 and −5.11).
      Risk Sentinel: This appears to be a dynamic risk management module that closed a significant number of trades. These exits were predominantly losses, suggesting the module is designed to cut trades before they hit a predefined stop-loss when market conditions turn unfavourable.
   3. GPO (General Pattern Orders): Exits like "M5_V_SHAPE" and "meta_failed" indicate a pattern-based logic is also at play.

3. Performance Trajectory: The account experienced a significant drawdown mid-week, particularly on October 7th, with the largest loss of -7.95. The week concluded with a highly volatile session on October 10th, where the system booked three large wins (7.95.The week concluded with a highly volatile session on October 10th, where the system booked three large wins (5.40, $6.78, 3.07) but also a significant loss(−3.07) but also a significant loss(−5.11), ultimately failing to recover the week's losses. A Profit Factor of 0.84 signifies that for every dollar of risk, the system only generated $0.84 in profit, confirming an unprofitable performance.
   

3.0 Performance Analysis: Week 2 (October 13 - 16, 2025)

This period marked a notable shift in performance. A key observation is the update of the trading algorithm, with comments changing from "GATE v10.x" to "GATE v11.0.x" starting October 14th.

3.1 Key Performance Metrics (Week 2)


Starting Balance: $174.57
Ending Balance: $187.57
Net Profit / Loss: +$13.00
Total Closed Trades: 35
Winning Trades: 20 (57.1%)
Losing Trades: 15 (42.9%)
Win Rate: 57.1%
Gross Profit: $40.82
Gross Loss: -$27.82
Profit Factor: 1.47
Average Win: $2.04
Average Loss: -$1.85
Largest Win: $6.25 (Oct 14)
Largest Loss: -$8.90 (Oct 14)

3.2 Narrative Analysis (Week 2)

1. Trading Activity: Trading frequency decreased significantly to 35 trades over four days, roughly half the volume of the previous week. This suggests the newer version of the algorithm may be more selective in its trade entries.
2. Strategy Behaviour:
   1. System Update: The clear transition to GATE v11.x is the most critical factor. This update appears to have directly influenced the trading outcomes.
   2. Risk vs. Reward: The average win of 2.04 is nearly double that of Week1 (2.04 is nearly double that of Week 1 (1.07), while the average loss ($1.85) remained proportionally higher but was overcome by the improved winning performance.
   3. Major Loss Event: The system incurred its single largest loss of the entire two-week period (-$8.90 on Oct 14). However, unlike in Week 1, the system recovered from this drawdown effectively. It immediately followed with a series of strong wins, including the week's largest gain of $6.25.

3. Performance Trajectory: The week started with a few losing trades but quickly found its footing after the system update. Despite the large "Risk Sentinel" loss on October 14th, the system consistently generated profits, leading to a steady equity curve growth for the remainder of the week. The Profit Factor of 1.47 indicates a healthy, profitable strategy where every dollar risked generated $1.47 in return.

4.0 Comparative Analysis & Conclusion

Week 1 (Oct 6-10)
Week 2 (Oct 13-16)

Net Profit / Loss:
Week 1:-$7.73
Week 2: +$13.00
Change: + $20.73

Profit Factor
Week 1: 0.84 (Unprofitable)
Week 2: 1.47 (Profitable)
Change: + 75%

Win Rate
Week 1: 52.1%
Week 2: 57.1%
Change: + 5.0%

Total Trades
Week 1: 71
Week 2: 35
Change: - 50.7%

Average Win$
Week 1: 1.07
Week 2: $2.04
Change: + 90.7%

Average Loss
Week 1:-$1.39
Week 2: -$1.85
Change: - 33.1%

Largest Loss
Week 1: -$7.95
Week 2: -$8.90
Change: - 11.9%

Key Findings:

1. Profitability: The primary difference is the shift from a net loss to a substantial net profit. Week 2 outperformed Week 1 by $20.73.
2. Strategy Efficacy (Profit Factor): The most significant improvement was the Profit Factor, which moved from an unprofitable 0.84 to a robust 1.47. This indicates the GATE v11.x system is fundamentally more effective at generating profits relative to its losses.
3. Trade Selectivity: The 50% reduction in trade frequency suggests GATE v11.x employs a more refined entry logic, avoiding the lower-quality setups that may have contributed to the "churn" and net loss in Week 1.
4. Reward-to-Risk Profile: While the average loss increased in Week 2, the average win nearly doubled. This improved reward profile was the key driver of profitability, as winning trades contributed more significantly to the bottom line, easily offsetting the losses.

Conclusion:
The performance data strongly suggests that the upgrade to the GATE v11.x algorithm was a successful and pivotal event. While Week 1's strategy was active but inefficient, Week 2's strategy was more selective, resilient, and ultimately profitable. It successfully navigated market volatility to achieve a positive return, recovering from a significant loss that the prior week's system may have struggled with.

So people are not confused, this is not the GATE SPX500. This is a supply and demand system that can 'generally' work with any symbol. 'GATE SPX500' I have been posting about is optimised for the SPX500 only.

1. Timeframe: 15-minute candles.
2. Trading Zone: Only during Prime Hours (07:00-11:00 UTC and 13:00-17:00 UTC).
3. Primary Signal: A "Failed Breakdown" pattern (for a long trade) or a "Failed Breakout" pattern (for a short trade) that meets all the quality criteria (preceded by a trend, good volatility, and a strong closing body).
4. Baseline Performance: An 81.3% win rate for a bullish move over the next 45 minutes.

Now, let's introduce two confirming signals that must occur on the same candle as the Primary Signal to qualify for a trade.

Confirming Signal #1: The Volume Spike

The "Why": Price action tells you what happened, but volume tells you the conviction behind it. A reversal pattern on low volume can be noise, but a reversal on a significant spike in volume indicates that institutional players have stepped in, forcefully rejecting the prior move and validating the new direction. This is one of the strongest forms of confirmation.

The Rule:
The TICKVOL of the 15-minute Trigger Candle must be at least 50% greater than the average TICKVOL of the previous 10 candles (a 2.5-hour lookback).
Statistical Impact:

Analysis of 32 instances of the "A++" Bullish Reversal pattern:

1. Of the 32 setups, 19 occurred with a confirming Volume Spike.
2. The performance of these 19 trades:
   1. Wins: 17
   2. Losses: 2
   3. Win Rate (Next 45 min): 89.5%

Conclusion: Adding the volume filter boosts the win rate from 81.3% to nearly 90%. It effectively filters out lower-conviction reversals, leaving you with setups backed by significant market participation.

Confirming Signal #2: The Structural Test

The "Why": The location of a pattern is paramount. A reversal signal is far more powerful if it occurs at a logical, technically significant price level. This shows that the market is respecting a previous area of support or resistance. Trading without regard to structure is like sailing without a map.
The Rule:

1. For a Bullish "Failed Breakdown": The LOW of the Trigger Candle must test (i.e., touch or briefly break) the LOW of the previous 1-hour range (the lowest point of the preceding four M15 candles).
2. For a Bearish "Failed Breakout": The HIGH of the Trigger Candle must test the HIGH of the previous 1-hour range.

This rule ensures you are buying at a logical area of support after sellers have failed, or selling at a logical area of resistance after buyers have failed.
Statistical Impact:

After further analysis of the "A++" Bullish Reversal pattern:

1. Of the 32 setups, 21 occurred after testing a key structural low.
2. The performance of these 21 trades:
   1. Wins: 18
   2. Losses: 3
   3. Win Rate (Next 45 min): 85.7%

Conclusion: The structural filter also provides a significant boost, increasing the win rate from 81.3% to 85.7%. It confirms that your entry is not random but is aligned with recent price behaviour.

The "Trifecta": Combining All Rules for the Ultimate A++ Setup

The highest-quality trade occurs when the Primary Signal is confirmed by both the Volume Spike and the Structural Test. This is the confluence of all ideal conditions.

1. Condition: "A++" Primary Signal + Volume Spike + Structural Test.

Final Analysis Results:

1. Total Occurrences (Bullish Trifecta): 11 times
2. Wins (Next 45 min): 10
3. Losses (Next 45 min): 1
4. Final Win Rate: 90.9%

While it appears infrequently (only 11 times in the provided data), its historical performance is exceptionally high.

Final Rules for Entry and Exit

Entry Rules (for a LONG trade):
You enter a long position if ALL of the following conditions are met:

1. Time: The current time is within a Prime Trading Zone (07:00-11:00 UTC or 13:00-17:00 UTC).
2. Context: The two previous M15 candles were both bearish (SS context), establishing a clear downward trend to reverse.
3. Primary Signal: The current M15 candle is a "Failed Breakdown" pattern (Low made in the first 10 mins, Close in the top third of the range).
   1. The candle's range must be > 6.50 points.
   2. The candle's body must be > 25% of its range.

4. Confirmation 1 (Volume): The candle's TICKVOL is > 1.5x the average of the last 10 candles.
5. Confirmation 2 (Structure): The LOW of the candle successfully tested the low of the previous hour.

Entry Point: Enter at the open of the next M15 candle.
Exit Rules:
A structured exit plan is critical to lock in profits and manage risk.

1. Stop Loss: Place your initial stop loss just below the low of the Trigger Candle. This defines your risk. If this level is breached, the thesis is invalidated.
2. Profit Target: Your primary profit target should be a 1.5x reward/risk ratio.
   1. Example: If your Trigger Candle's low is at 6650 and you enter at 6655, your risk is 5 points. Your profit target would be 6655 + (5 * 1.5) = 6662.50.

3. Time-Based Exit: If neither your Stop Loss nor Profit Target is hit within the next three M15 candles (45 minutes), exit the trade at the market price. Our analysis showed the pattern's predictive power is strongest within this window, so if the move hasn't materialized by then, it's best to exit and preserve capital.

# --- v1.12: The Go/No-Go Checklist Integration ---
# 1. STRATEGIST AI PROMPT OVERHAUL: The AI's prompt (`get_strategist_prompt`) has
# been rebuilt to be a strict "Go/No-Go Checklist" that directly mirrors the
# user-defined 7-point trading plan.
# 2. EXPLICIT VALIDATION: The AI is now required to sequentially answer and validate
# each of the core checklist questions in its rationale before making a decision:
# - 1. Session Check
# - 2. Market Structure Check
# - 3. S/D Zone Identification
# - 4. Zone Test Confirmation
# - 5. Signal Candle Confirmation
# - 6. Risk-to-Reward Assessment
# 3. ENHANCED CONTEXT PASSING: The `detect_trifecta_pattern` function now explicitly
# passes the preceding trend direction to the AI, enabling a more accurate
# assessment of the "Market Structure" checklist item.
# 4. AI ROLE CLARIFICATION: The AI's task is now unambiguously to act as the final
# gatekeeper for the 7-point plan. It confirms the mechanical signal and the
# trade environment against a human-centric checklist before execution. This
# makes its reasoning transparent and directly aligned with the strategy.
#
# --- v1.11: Enhanced AI Confirmation Logic ---
# 1. REFINED STRATEGIST PROMPT: Upgraded to include a more nuanced "Strategist's
# Confirmation Checklist" for better AI validation.
# 2. AI VALIDATION OF TRIFECTA: Tasked the AI with re-validating the core components
# of the detected Trifecta signal.
#
# --- v1.10: Trifecta Strategy & Market Session Intelligence ---
# 1. CORE STRATEGY ENHANCED: Integrated the "Trifecta" pattern.
# 2. TRIFECTA PATTERN ENGINE: Implemented a mechanical engine to detect the A++ setup.
# 3. MARKET SESSION MANAGER: Introduced a class to provide temporal context and
# define trading modes (PRIME_SESSION, CAUTIOUS_TRANSITION, LULL_AVOID).
#
# --- v1.00: Price Action & Market Structure Engine ---
# 1. CORE STRATEGY REBUILT: Aligned with a day trading plan based on pure price
# action, market structure, and Supply/Demand zones.

1. Time of Entry: 2025-10-17 19:10:04 UTC (15:10 NY Time)
2. Action: SELL at a price of 6685.13.
3. AI Confidence: 0.95 (Extremely High).
4. AI's Primary Driver: "A prior Bullish V-shape has failed by breaking its pivot low. This is a high-probability 'bull trap' setup, signalling a strong rejection of higher prices."
5. Signal Source: This was triggered by the meta_failed_bullish_v pattern in detector_patterns.json, which was assigned a "CRITICAL" priority.

2. Market Context at the Time of Entry
While the AI saw a valid M5 pattern, it critically misinterpreted or ignored the following higher-timeframe data:

1. M15 Intraday Trend: The M15 trend was unambiguously bullish. The fast EMA was above the slow EMA, and the price was trading above both, indicating consistent buying pressure throughout the session.
2. Junior Analyst Strategic Bias: The M15 Strategist (Junior AI) had correctly identified the bullish structure and had likely proposed an UP or FLAT bias, which the Senior AI overrode.
3. Market Heartbeat: The trade occurred during the AFTERNOON_COOLDOWN phase (HB_COOLDOWN), a period where strong, impulsive breakouts are less likely to follow through.

3. The Logical Flaw: A Failure of Hierarchy
The loss was not caused by a faulty pattern detector; the "bull trap" pattern was real. The failure was in the system's rules of engagement.

1. Absolute Override Command: The action_guidance for the meta_failed_bullish_v pattern was flawed. It stated, "This meta-pattern should override conflicting standard signals."
2. Misinterpretation by AI: The Senior AI interpreted this as an absolute command to ignore all other data points, including the most critical piece of intraday context: the M15 trend.
3. Resulting Action: Faced with a CRITICAL signal that told it to override everything else, the AI executed the SELL order with high conviction, effectively betting against the dominant intraday momentum. The market then continued its bullish path, stopping the trade out for a loss, a predictable outcome when fighting an established trend.

This trade is a textbook example of a Hierarchy Violation, where a low-timeframe tactical signal was incorrectly prioritized over high-timeframe strategic context.

System Corrections
The following specific, targeted changes have been made to the system's logic and AI instructions to permanently fix this flaw.
1. Hardened Senior AI "Conflict Resolution Protocol"
The core instructions in the get_senior_prompt function have been fundamentally rewritten to enforce a non-negotiable decision hierarchy. The AI is no longer just given data; it is given explicit rules on how to resolve conflicts.
New Mandatory Protocol added to the prompt:

--- **CONFLICT RESOLUTION PROTOCOL (MANDATORY)** --- Your core task is to resolve conflicts between signals. Intraday trend context (M15) ALWAYS wins over lower-timeframe patterns (M5). - **SCENARIO:** If the M15 trend is clearly BULLISH (e.g., price is consistently above M15 EMAs, ADX > 25) and a counter-trend `BEARISH_V_REVERSAL_SETUP` (M5 pattern) appears, you MUST VETO the SELL signal. - **REQUIRED RATIONALE:** If you veto, your rationale must state: "Vetoing M5 bearish pattern due to direct conflict with established M15 bullish intraday trend." - **META-PATTERN EXCEPTION:** A "meta-pattern" (like a Bear Trap) is the ONLY signal that can challenge an M15 trend, but you must still explicitly state in your rationale that you are overriding the M15 trend due to the meta-pattern's higher conviction, and only if the M15 trend is not excessively strong.

1. How this Prevents the Failure: Under this new protocol, the Senior AI would have been forced to acknowledge the conflict between the bearish M5 pattern and the bullish M15 trend. It would have identified that the trend was not weak and, following its new primary directive, would have issued a HOLD decision, explicitly stating the reason for the veto. The trade would have been prevented.

2. Refined Detector Pattern Guidance (detector_patterns.json)
The absolute "override" command has been removed from the action_guidance of critical meta-patterns and replaced with more nuanced instructions that empower the AI to make a smarter, context-aware decision.
For the meta_failed_bullish_v pattern:

1. Old Guidance: "This meta-pattern should override conflicting standard signals."
2. New Guidance (v11.3.0): "This is a high-probability 'bull trap' SELL signal. **Vigorously prioritize this signal, BUT it must be validated against the M15 trend.** Veto the trade if the M15 trend is strongly and clearly bullish. An override is only authorized if the M15 trend is flat or already bearish."
3. How this Prevents the Failure: This change provides the AI with the necessary nuance. It confirms the pattern's importance but explicitly links its execution to the M15 context. Instead of a blind override, the AI is now instructed to perform the exact validation check that was missing on October 17th.

Conclusion
The October 17th loss was a valuable diagnostic event that exposed a fundamental weakness in the system's logic. By implementing a stricter decision-making hierarchy in the Senior AI's core programming and refining the guidance of our most powerful signals, the system is now far more robust. It is equipped not just to identify patterns, but to correctly interpret their relevance within the broader market context, preventing it from taking high-risk, counter-trend trades and significantly improving its overall reliability.


SPX500 Parallel Test Executive Summary
On October 17, 2025, the trading system experienced a series of significant, preventable losses. A critical post-mortem analysis revealed that these failures were not random but stemmed from three core areas:

1. Critical Logic Flaws: The system fundamentally misinterpreted its own signals, most notably executing a BUY order on a high-conviction bearish pattern.
2. Code & Execution Bugs: Multiple trades were either missed or improperly managed due to software bugs, including invalid stop-loss calculations and a flawed risk-tightening mechanism.
3. Environmental Mismanagement: The system failed to adequately account for the extreme volatility of the New York open, leading to an immediate and substantial loss.

This report will dissect each major failure from the October 17th session and detail the specific code and logic changes in version v11.1.0 that directly address these root causes. The corrections are targeted, comprehensive, and are expected to significantly improve system robustness and profitability.
2. Root Cause Analysis of Key Failures (October 17th)
A. Failure of Strategic Hierarchy: The Contradictory Buy Trade (#201956606)

1. Incident: At 01:40:06, the Detector Engine correctly identified a meta_failed_bullish_v pattern—a high-probability "bull trap" which is a strong SELL signal. However, the system's strategic bias was BULL_TREND_DIP_HUNTING. Instead of prioritizing the critical meta-pattern, the Senior AI misinterpreted it as a "dip entry opportunity" and executed a BUY order. The market proceeded to sell off as the pattern predicted, resulting in a loss of -$7.25.
2. Root Cause: A fatal flaw in the AI's decision-making hierarchy. The broad, hypothesis-driven "strategic bias" (Junior AI's role) was allowed to override a specific, high-conviction, real-time price action signal (the meta-pattern).
3. Correction in v11.1.0:
   1. Empowered Senior AI: The Senior AI's master prompt has been fundamentally rewritten to establish a strict decision hierarchy: JFX Price Action > Meta-Patterns > Other Signals > Junior AI Bias.
   2. New Prompt Logic: The new prompt explicitly instructs the Senior AI: "Treat the Junior Analyst proposal... as a strategic bias or hypothesis. Use the live JFX context, Pattern Alerts, and M5 Candle DNA to confirm or invalidate that bias for an immediate entry." and "Conflict Resolution: meta-patterns (like a Bull/Bear Trap) are your primary tool for conflict resolution." This ensures a bull trap is always treated as a sell signal, regardless of the underlying bias.

B. Failure to Manage NY Open Volatility: The Immediate Stop-Out (#202082787)

1. Incident: At 23:40:07 SYD (09:40 NY), just 10 minutes after the system's "quiet zone" ended, a trade was executed based on a pattern formed during the chaotic opening moments of the New York session. The market's extreme volatility immediately stopped the trade out for a -$12.02 loss.
2. Root Cause: The PRE_EVENT_CAUTION state was too brief and ended too soon. The system was not adequately shielded from the predictable and often irrational volatility of the first 15-30 minutes of the NY open.
3. Correction in v11.1.0:
   1. Implemented "Quiet Zone": A hard-coded "PRE_EVENT_CAUTION" state is now active from 9:15-9:30 AM NY time. The code logic if session == "PRE_EVENT_CAUTION": ... continue now strictly prohibits any new trade analysis or execution during this window, preventing entries based on pre-market noise.

C. Critical Execution & Management Bugs

1. Incident 1 - Invalid Stops (Retcode 10016): On three separate occasions (11:55:08, 15:00:09, 15:05:08), the system attempted to open a trade with a stop-loss that was too close to the entry price (e.g., 0.38 points). The broker rejected these orders as invalid, resulting in missed opportunities.
2. Root Cause: The stop-loss calculation logic, particularly for pattern-based SLs, lacked a minimum threshold. It could calculate a mathematically correct but practically impossible stop-loss.
3. Correction in v11.1.0:
   1. Fixed Order Execution Failures: The open_trade function now includes a pre-send check against MINIMUM_PROTECTIVE_SL_POINTS and the broker's minimum distance requirements.
   2. Robust Calculations: Fallbacks have been added to ATR and StdDev calculations to ensure they never return zero, preventing division-by-zero errors that could lead to invalid stop calculations.

4. Incident 2 - Flawed Risk Sentinel (Trade #201975306): On a losing trade, the Risk Sentinel attempted to "tighten" the stop-loss to a price that the market had already passed, causing the modification to be rejected by the broker and failing to reduce risk.
5. Root Cause: The defensive logic was flawed. It was programmed to move the stop to a fixed point (like breakeven) which is impossible if the price is already further in loss.
6. Correction in v11.1.0:
   1. Corrected Risk Sentinel Logic: The "Defensive" tier logic was completely rewritten. It now calculates a new stop-loss by moving it halfway from its original position to the entry price. This guarantees a valid stop placement while still effectively cutting the remaining risk by 50%.

D. Signal Quality and Whipsaw Vulnerability

1. Incident: Multiple trades, such as #201962692, were based on valid patterns that quickly failed due to the choppy, mean-reverting market conditions prevalent on October 17th (Hurst exponent was consistently low). The system was too sensitive to lower-quality V-shape patterns.
2. Root Cause: The conviction threshold for standard V-shape patterns was too low, allowing noisy signals to trigger the system's logic in an unfavourable, whipsaw-prone environment.
3. Correction in v11.1.0:
   1. Deprioritized Noisy V-Shapes: The default min_conviction_score for M5_V_SHAPE_BULL and M5_V_SHAPE_BEAR patterns in detector_patterns.json has been increased from 5.0 to 7.5.
   2. Refined Stalk & Pounce: The confirmation logic for V-shapes is no longer a simple directional close. It now requires a more robust "impulsive" candle (body > 70% of range) or an RSI cross, filtering out weak confirmations.
   3. Enhanced Playbook Logic: The audit_bull_sharp_dip_buy_v1 playbook was updated to require a price_in_jfx_demand_zone condition, adding an essential layer of structural confirmation to prevent chasing dips in open space.

3. Conclusion
The losses on October 17th were a direct result of identifiable and correctable flaws. The v11.1.0 update is not a minor tweak but a significant architectural and strategic overhaul.

1. The AI's decision hierarchy is now clear and logical, preventing strategic biases from overriding critical price action signals.
2. Critical code bugs related to order execution and risk management have been fixed, ensuring the system can now act on its decisions reliably.
3. Environmental protections, such as the pre-NY "Quiet Zone," have been implemented to shield the system from predictable periods of extreme and irrational volatility.

By addressing these core failings the update is fundamentally more robust, intelligent, and risk-aware. It is now better equipped to capitalize on its core strength—identifying high-probability patterns—while being better protected against the specific conditions and errors that caused the prior losses.

GATE TEST UPDATES

1. TEST 2: The bot would close all trades on Friday and enter "Weekend Sleep" mode. It had no capacity to trade the weekend gap.
2. TEST: A completely new, self-contained strategy has been added to capitalize on the price gap between Friday's close and Monday's open. It operates in three distinct phases:
   1. Friday Setup: Enters a BUY position in the final moments of trading on Friday, based on a "Last Hour Pullback" pattern.
   2. Monday Analysis: Before the CFD market opens, it analyses S&P futures, Nikkei, and AUD/JPY to gauge overnight sentiment and decide whether to aim for a large or small profit.
   3. Monday Execution: Manages the final exit as soon as the market opens based on the analysis.

Daily Maintenance Window:

1. BOTH HAVE: A dedicated MAINTENANCE_HALT state is now part of the session controller. Every day from 17:30 to 18:30 NY Time, the system will automatically:
   1. Close any open trades.
   2. Halt new trading.
   3. Run a maintenance script to rebuild its Markov transition matrix and perform other learning tasks.
   4. Resume trading after the window closes.

1. Stallone has his tech-head weed employee hack the account of his rival and steal his money, then send some money, I remember it was $1M this to the FBI woman he was seeing who was shot in the previous episode and is in hospital recovering.
   1. Whilst the FBI were access the account at the same time they see the money disappear in front of them and the main rival sees that too and goes ape...

2. Stallone and his group have a big bar fight with the main rival and their group where the main rival character is killed off when Stallone skewers him on a mounted animal on a pillar in the bar.
   1. The young black side kick gets shot.

3. Final scenes of the episode there is a celebration in the bar, Stallone has his casino licence, he goes outside and gets arrested in front of his daughter on charges of trying to bribe the FBI woman - which was connected to the hack and he obviously was doing for love and not for legal judo.