|
no
esperienza
|
5
prodotti
|
48
versioni demo
|
|
0
lavori
|
1
segnali
|
0
iscritti
|
I'm m a Forex en stocks trader en software developer
Florin Tudor
After 1 week whit the same settings King of Gold AI Agent have a good 👍 score 20% profit which is ok with moderate settings ! Important is to keep that winings in the account en grow furthermore!The AGENT learn new strategy alone en save to .CSV files all the patterns en volume analizes en menny more !
· 4
Mostra tutti i commenti (4)
Florin Tudor
2026.03.31
I think because OF TESLA en XAU en forex
Florin Tudor
2026.04.01
En 12 other 📉 charts
zambilica
2026.04.01
Woow The King 👑
Florin Tudor
For my clients!! Because I changed the name on Queen of Gold AI en King of Gold AI the existing clients can not update so I will let the price to be minimum so you can purchase en get the latest version!!
· 3
Florin Tudor
Prodotto pubblicato
Queen of Gold AI UCB Agents è un Expert Advisor (EA) MQL5 adattivo che combina una logica di ingresso basata su regole con un livello bandito Upper Confidence Bound (UCB) contestuale. Il suo scopo non è solo rilevare opportunità di trading, ma anche apprendere quale azione sia più efficace in un dato contesto di mercato e conservare questa conoscenza tra le sessioni. Funzionamento L’EA parte da un motore di segnali deterministico. Il trigger di base si basa sul comportamento delle candele
· 4
Florin Tudor
Prodotto pubblicato
L’Agente King of Gold AI UCB è un avanzato algoritmo di apprendimento automatico progettato per ottimizzare le decisioni di trading attraverso apprendimento adattivo e selezione probabilistica. Caratteristiche principali: Bandito decisionale UCB con filtraggio contestuale e ottimizzazione dei segnali Architettura a triplo bandito (livelli Decisione, Esecuzione, Profilo) Dimensionamento dinamico dei lotti con gestione del rischio adattiva Memorizzazione persistente del contesto e inizializzazione
· 4
Florin Tudor
Prodotto pubblicato
Agente AI Multi-Arm Bandit – Motore di Apprendimento Adattivo e Decisionale Un sistema di trading avanzato, basato su un algoritmo multi-arm bandit, che apprende e si adatta continuamente alle condizioni di mercato. L’agente ottimizza in tempo reale l’esecuzione delle operazioni, la dimensione delle posizioni e le decisioni sulla base del feedback sulle performance. Logica di Decisione Utilizza la formula UCB: exploit + explore, dove exploration = C * sqrt(log(totalPulls + 1) / armPulls)
· 4
Florin Tudor
Maximum scalper AI 🔍 File Overview
UCB_STATE_V4 = a snapshot of your bandit algorithm’s state:
Which combinations it has tried
How many times it tried them
The profit/reward for each
Its internal scoring
Basically, it’s the memory of your AI.
🧠 Structure
1️⃣ Header
UCB_STATE_V4
1
→ Version + config info.
2️⃣ Default/Fallback Strategy
-1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY
-1 probably = SELL
AGGRESSIVE + GREEDY = style
Trades = 0, Profit = 0 → never used
3️⃣ Important strategies that were actually used
1|1|1|1|1|0|2|0|0|NORMAL|MODERATE|EXEC
96 trades
Profit: 37.42
Score: 0.40
🔥 This is your main winning strategy
Chosen 96 times
Positive profit and positive score → bandit prefers this
-1|1|1|1|1|0|2|0|0|NORMAL|MODERATE|EXEC
22 trades
Profit: 13.32
Score: negative
Positive profit but negative score → probably high drawdown or risk
4️⃣ Other strategies
0 trades
0 profit
Not explored yet or never chosen
5️⃣ Symbol statistics
EURUSD
118 trades
Win rate ~55%
The bandit is learning, not random
6️⃣ Decision history
timestamp | chosen strategy | reward
Shows when a strategy was chosen and the outcome
❗ Key Problem (why it’s not working in Tester)
Strategy Tester in MT5 runs in a sandbox
DLL cannot write anywhere freely
It only writes to:
MQL5\Files\
Tester\Agent-xxx\MQL5\Files\
✅ Live accounts don’t have this restriction → that’s why it works there.
🔥 Recommended Testing Environments
Demo account – works like live, writes files normally
Forward testing on chart – watch bandit evolution
VPS – run continuously, bandit keeps learning
Conclusion:
Your bandit is working (96 trades + scores)
The algorithm is fine
The limitation is MT5 Strategy Tester, not the AI itself
UCB_STATE_V4
1
-1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY 0 1774229303 0 0.00000000 0.00000000 0 0.00000000 0.00000000 0 0.00000000 0.00000000 0 0.00000000 0.00000000
3
1|1|1|1|1|0|2|0|0|NORMAL|MODERATE|EXEC 96 1774229084 96 37.42000000 0.40015989 0 0.00000000 0.00000000
-1|1|1|1|1|0|2|0|0|NORMAL|MODERATE|EXEC 22 1774229084 22 13.32000000 -0.16765199 0 0.00000000 0.00000000
-1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY|EXEC 0 1774229303 0 0.00000000 0.00000000 0 0.00000000 0.00000000
6
0|1|0|0|2|0|0|AGGRESSIVE|GREEDY 0 1774229304 0 0.00000000 0.00000000 0 0.00000000 0.00000000
0|2|0|0|2|0|0|AGGRESSIVE|GREEDY 0 1774229363 0 0.00000000 0.00000000 0 0.00000000 0.00000000
0|2|0|0|2|0|0|AGGRESSIVE|CONSERVATIVE 0 1774229408 0 0.00000000 0.00000000 0 0.00000000 0.00000000
0|2|0|0|2|0|0|AGGRESSIVE|MODERATE 0 1774229415 0 0.00000000 0.00000000 0 0.00000000 0.00000000
0|1|2|0|1|2|0|AGGRESSIVE|CONSERVATIVE 0 1774596950 0 0.00000000 0.00000000 0 0.00000000 0.00000000
0|1|2|1|1|2|0|AGGRESSIVE|CONSERVATIVE 0 1774597077 0 0.00000000 0.00000000 0 0.00000000 0.00000000
1
EURUSD 1 118 82 0.55416158 0.12497833 50.00000000
2
EURUSD 1 1 1 1 1 0 96 66 37.42000000 -16.84000000
EURUSD 1 -1 1 1 1 0 22 16 13.32000000 25.44000000
0
2
1547155986 -1 -1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY 1 5.00000000 -1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY|EXEC 0 1774229304
1547159109 -1 -1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY 1 5.00000000 -1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY|EXEC 0 1774229361
UCB_STATE_V4 = a snapshot of your bandit algorithm’s state:
Which combinations it has tried
How many times it tried them
The profit/reward for each
Its internal scoring
Basically, it’s the memory of your AI.
🧠 Structure
1️⃣ Header
UCB_STATE_V4
1
→ Version + config info.
2️⃣ Default/Fallback Strategy
-1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY
-1 probably = SELL
AGGRESSIVE + GREEDY = style
Trades = 0, Profit = 0 → never used
3️⃣ Important strategies that were actually used
1|1|1|1|1|0|2|0|0|NORMAL|MODERATE|EXEC
96 trades
Profit: 37.42
Score: 0.40
🔥 This is your main winning strategy
Chosen 96 times
Positive profit and positive score → bandit prefers this
-1|1|1|1|1|0|2|0|0|NORMAL|MODERATE|EXEC
22 trades
Profit: 13.32
Score: negative
Positive profit but negative score → probably high drawdown or risk
4️⃣ Other strategies
0 trades
0 profit
Not explored yet or never chosen
5️⃣ Symbol statistics
EURUSD
118 trades
Win rate ~55%
The bandit is learning, not random
6️⃣ Decision history
timestamp | chosen strategy | reward
Shows when a strategy was chosen and the outcome
❗ Key Problem (why it’s not working in Tester)
Strategy Tester in MT5 runs in a sandbox
DLL cannot write anywhere freely
It only writes to:
MQL5\Files\
Tester\Agent-xxx\MQL5\Files\
✅ Live accounts don’t have this restriction → that’s why it works there.
🔥 Recommended Testing Environments
Demo account – works like live, writes files normally
Forward testing on chart – watch bandit evolution
VPS – run continuously, bandit keeps learning
Conclusion:
Your bandit is working (96 trades + scores)
The algorithm is fine
The limitation is MT5 Strategy Tester, not the AI itself
UCB_STATE_V4
1
-1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY 0 1774229303 0 0.00000000 0.00000000 0 0.00000000 0.00000000 0 0.00000000 0.00000000 0 0.00000000 0.00000000
3
1|1|1|1|1|0|2|0|0|NORMAL|MODERATE|EXEC 96 1774229084 96 37.42000000 0.40015989 0 0.00000000 0.00000000
-1|1|1|1|1|0|2|0|0|NORMAL|MODERATE|EXEC 22 1774229084 22 13.32000000 -0.16765199 0 0.00000000 0.00000000
-1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY|EXEC 0 1774229303 0 0.00000000 0.00000000 0 0.00000000 0.00000000
6
0|1|0|0|2|0|0|AGGRESSIVE|GREEDY 0 1774229304 0 0.00000000 0.00000000 0 0.00000000 0.00000000
0|2|0|0|2|0|0|AGGRESSIVE|GREEDY 0 1774229363 0 0.00000000 0.00000000 0 0.00000000 0.00000000
0|2|0|0|2|0|0|AGGRESSIVE|CONSERVATIVE 0 1774229408 0 0.00000000 0.00000000 0 0.00000000 0.00000000
0|2|0|0|2|0|0|AGGRESSIVE|MODERATE 0 1774229415 0 0.00000000 0.00000000 0 0.00000000 0.00000000
0|1|2|0|1|2|0|AGGRESSIVE|CONSERVATIVE 0 1774596950 0 0.00000000 0.00000000 0 0.00000000 0.00000000
0|1|2|1|1|2|0|AGGRESSIVE|CONSERVATIVE 0 1774597077 0 0.00000000 0.00000000 0 0.00000000 0.00000000
1
EURUSD 1 118 82 0.55416158 0.12497833 50.00000000
2
EURUSD 1 1 1 1 1 0 96 66 37.42000000 -16.84000000
EURUSD 1 -1 1 1 1 0 22 16 13.32000000 25.44000000
0
2
1547155986 -1 -1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY 1 5.00000000 -1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY|EXEC 0 1774229304
1547159109 -1 -1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY 1 5.00000000 -1|2|0|0|0|0|2|0|0|AGGRESSIVE|GREEDY|EXEC 0 1774229361
· 3
Florin Tudor
Wat is this ? MQL5 spam rewiew from clients?ho else get this from MQL5 WEB?is not normal to manipulate the clients
· 2
Florin Tudor
Aa wait for the deep but the agents will go whit the trend ,the gift 🎁 is verry advance so my clients will receive this multi arm bandit agents EA for Free!
· 2
Florin Tudor
My Agents are very bizzy:)) the market is very choppy:)) but no problem sir we must make 🤑💰 for master Tudor:)
· 3
Florin Tudor
inside the King of gold en Maximum Scalper the workers the Ai Agents are working day en night ,the file/data folder/mql5/Files is where you will see the agents keep the information en model also wat simbol they work on ,enjoy this is made by a trader for traders!!
· 3
Florin Tudor
🎁 Maximum Scalper In 1 hours ,One day free Download for all clients to be able to tested out on Demo or Cent account!after 2 days the price will come back on ! Test the Agents en if you find something Rong just message me !
· 3
Florin Tudor
🎁 Maximum Scalper In 2 hours One day free Download for all clients to be able to tested out on Demo or Cent account!
I will make some set.files in comments!!
I will make some set.files in comments!!
· 3
Florin Tudor
Maximum Scalper AI has evolved from a simple rule-based reversal scalper into a locally adaptive MQL5 Expert Advisor with persistent AI-driven decision logic.
This upgrade introduces a self-learning multi-layer bandit system that adapts directly from real trade outcomes.the agent is build in so it will make his files for every simbol is trading in your mql5 files data folder
This upgrade introduces a self-learning multi-layer bandit system that adapts directly from real trade outcomes.the agent is build in so it will make his files for every simbol is trading in your mql5 files data folder
· 2
Florin Tudor
Maximum Scalper AI has been upgraded from a rule-based reversal scalper into a locally adaptive MQL5 expert advisor with persistent AI-style decision memory.
The update adds a multi-layer local bandit framework that learns directly from trade outcomes without any DLL dependency. The EA now uses:
a UCB decision bandit for HOLD / SMALL / NORMAL / STRONG
an execution bandit for MARKET / LIMIT
an auto-close bandit for DIRECT_CLOSE / TRAIL
Its learning context is no longer simplistic. It now evaluates signals using:
consecutive bullish/bearish bar structure
volume regime and volume ratio
volume profile bias
spread conditions
session timing
adaptive performance mode
greed/conservative signal state
equity vs balance condition
floating drawdown stress
safe mode state
The EA also gained persistent on-disk learning memory, so it can retain what it has learned across restarts. It stores:
UCB trade-decision memory
execution memory
auto-close memory
symbol/timeframe performance memory
setup-pattern statistics
active and pending contextual link
The update adds a multi-layer local bandit framework that learns directly from trade outcomes without any DLL dependency. The EA now uses:
a UCB decision bandit for HOLD / SMALL / NORMAL / STRONG
an execution bandit for MARKET / LIMIT
an auto-close bandit for DIRECT_CLOSE / TRAIL
Its learning context is no longer simplistic. It now evaluates signals using:
consecutive bullish/bearish bar structure
volume regime and volume ratio
volume profile bias
spread conditions
session timing
adaptive performance mode
greed/conservative signal state
equity vs balance condition
floating drawdown stress
safe mode state
The EA also gained persistent on-disk learning memory, so it can retain what it has learned across restarts. It stores:
UCB trade-decision memory
execution memory
auto-close memory
symbol/timeframe performance memory
setup-pattern statistics
active and pending contextual link
· 3
Florin Tudor
The King of Gold.mq5 file (version 9.08) has been integrated with UCB (Upper Confidence Bound) Bandit AI - an advanced machine learning algorithm that optimizes trading decisions.
Core Features
UCB Decision Bandit with contextual filtering and signal boosting
Triple-Bandit Architecture (Decision, Execution, Profile bandits)
Dynamic lot multipliers and smart risk management
Persistent context savings and trade history warm-up
Complete learning telemetry
Core Features
UCB Decision Bandit with contextual filtering and signal boosting
Triple-Bandit Architecture (Decision, Execution, Profile bandits)
Dynamic lot multipliers and smart risk management
Persistent context savings and trade history warm-up
Complete learning telemetry
· 3
Florin Tudor
On BTC the new version of King of Gold AI performing verry well in first day of testing
· 2