production: 3 services running

Strategies for
algorithmic trading

Discovery, testing and verification of trading strategies on tick data. Custom framework, 20+ metrics per minute, lookahead bias protection. 6 years in trading, 6,000+ hours of R&D.

submit_request() Backtest examples

years in trading

6,000+

hours of R&D

years in IT

20+

metrics per minute

// background

Background

12 years in IT and product development. Worked in roles from engineer to product manager. Core skill — systematic approach to hypothesis verification and data-driven decision making.

6 years in financial analysis: parsing and analyzing brokerage reports, portfolio auditing, building tools for objective trading performance evaluation.

6 years in trading. Last 1.7 years — full immersion in algorithmic trading: 6,000+ hours of testing strategies, indicators and methodologies.

Methodologies: Smart Money Concepts, technical analysis, hybrid approach with probabilistic modeling based on quantiles and feature engineering.

career_path.timeline()

2014–2026 · 12 years

IT & Product

Engineering, product analytics, hypothesis validation

2020–2026 · 6 years

Trading

Smart Money, TA, manual trading on stock and crypto markets

2020–2026 · 6 years

Financial Analysis

Brokerage reports, portfolio audits, evaluation tools

2024–now · 6,000+ hours

Algorithmic Trading

Custom infrastructure, backtest framework, production bots

// rationale

Why I do this

The infrastructure is built and running: server, PostgreSQL + TimescaleDB, backtest framework, trading bots in production. Bots are autonomous — between iterations there's free time and a ready-made tool that can be used to help others.

Every new task is a reason to add new metrics, expand framework capabilities, add indicators that weren't there before. Your task makes my tool better.

Every component of the system is verifiable: data, code, test results. These aren't TradingView screenshots — this is working infrastructure that can be demonstrated.

// market_structure

BTC Microstructure

BTC pricing is driven by BlackRock through Aladdin — an algorithmic system controlling $10+ trillion in assets. Aladdin shapes liquidity and determines market movement.

Classic technical analysis — patterns, support/resistance levels, textbook RSI/MACD — is not part of these algorithms. The market systematically exploits the predictability of retail traders using these methods.

The only working approach: statistical hypothesis verification on data. Not finding patterns visually, but discovering statistically significant regularities in market microstructure: order flow, volume divergences, market maker behavior.

methodology = "verify" if hypothesis else "discard"

// tech_stack

Infrastructure

Production-grade stack. Not Jupyter scripts — a full-fledged system.

data_layer

Data

$aggTrades API — every trade from Binance Futures
$PostgreSQL 16 + TimescaleDB 2.24
$Minute aggregates: buy/sell qty, divergences, normalization
$Incremental update every 10 sec

backtest_engine

Framework

$Python: numpy, pandas, psycopg2
$Extensible: plug in any indicators
$Lookahead bias protection at architecture level
$Feature engineering: 7+ derived metrics

execution

Execution

$Binance Futures API: MARKET + TP/SL orders
$systemd services, auto-reconnect
$Flask dashboard: health, logs, stats
$Telegram: entry/exit alerts + P&L

$ pipeline --show

// backtest_framework

Framework internals

Fragment of real configuration and system output. Parameters altered.

strategy_config.py

# Strategy: Gated V2 — divergence crossing
STRATEGY = {
"lookback": 60,  # minutes
"tp_pct": 0.028,  # take profit 2.8%
"sl_pct": 0.018,  # stop loss 1.8%
"decision_mode": "exclusive",
"max_positions": 1,
}
FEATURES = [
"last_diff",     # norm_price - norm_diff
"slope_diff",    # linreg slope of divergence
"slope_np",      # price momentum
"slope_nd",      # volume momentum
"mean_diff",     # avg divergence
"price_vol",     # stddev of returns
"price_drift",   # mean of returns
]
LONG_GATE = {
"last_diff":  "<= Q75",
"slope_diff": ">= Q25",
"slope_np":   ">= Q25",
"slope_nd":   "<= Q25",
}

$ backtest --run --verbose

[2024-09-01 00:00] Loading data...
[2024-09-01 00:00] Period: 2024-08-25 → 2024-09-01
[2024-09-01 00:00] Rows loaded: 10,080 minutes
[2024-09-01 00:00] Scanning crossings...
[2024-09-01 00:01] Found 47 crossing points
[2024-09-01 00:01] Applying gate filters...
[2024-09-01 00:01] Passed LONG gate: 12/23
[2024-09-01 00:01] Passed SHORT gate: 8/24
[2024-09-01 00:01] Conflicts (both gates): 3
[2024-09-01 00:01] Exclusive mode: discarded
[2024-09-01 00:02] Simulating positions...
[2024-09-01 00:02] i_ptr: sequential, no overlap
[2024-09-01 00:03] Backtest complete
═══ Results ═══════════════════
Trades:        17
Win rate:      ██████░░░░ 58.8%
Profit factor: 1.42
Max drawdown:  -3.2%
Sharpe (ann.): 1.18
* parameters altered for demo

// backtest_results

Results: good and bad

Strategies vary. The goal is to separate working ones from non-working before risking capital.

Strategy A — Gated Divergence

BTCUSDC · 1min · Aug 2024

+12.4%

cumulative P&L

58.8%

win rate

1.42

profit factor

-3.2%

max DD

1.18

sharpe

Strategy B — RSI + MA Cross

BTCUSDC · 1min · Aug 2024

-8.7%

cumulative P&L

38.5%

win rate

0.72

profit factor

-11.3%

max DD

-0.45

sharpe

parameter_optimization

TP/SL parameter optimization: 36 combinations tested

Heatmap: profit factor for different TP/SL pairs. Green = profitable. Red = unprofitable.

TP \ SL	0.8%	1.2%	1.5%	1.8%	2.2%	2.8%
1.0%	0.65	0.71	0.82	0.88	0.91	0.85
1.5%	0.78	0.89	0.98	1.05	1.12	1.08
2.0%	0.85	0.96	1.10	1.22	1.28	1.18
2.5%	0.82	0.95	1.15	1.38	1.32	1.25
2.8%	0.80	0.92	1.12	1.42	1.30	1.20
3.5%	0.70	0.82	0.98	1.15	1.10	0.95

* profit factor. Green border = optimal zone. Values altered for demonstration.

// research_log

Conducted research

List of tests. Strategy details are not disclosed.

#01 · strategy

Gated V2 — Divergence Crossing

Price/volume divergence + 7-feature gate system + quantile thresholds

#02 · detection

Crossing Detection Algorithm

Intersection of normalized price and volume divergence curves

#03 · features

Feature Engineering Pipeline

7 derived metrics: slopes (linreg), volatility (std), drift (mean returns)

#04 · filtering

Quantile Gate System

Adaptive Q25/Q75 thresholds on a rolling 60-minute window

#05 · logic

Exclusive Decision Mode

Conflict filtering: signal only when one gate fires unambiguously

#06 · comparison

Local vs Global Quantiles

A/B test: adaptive quantiles (per-crossing) vs global (full dataset)

#07 · validation

Data Integrity Checks

Data completeness: minute gaps, aggregate correctness, normalization bounds

#08 · verification

Spec Compliance Testing

Automated verification: parameters, formulas, logic vs specification

#09 · e2e

End-to-End Integration

Full pipeline test: crossing -> features -> gate -> signal -> position -> exit

#10 · optimization

TP/SL Grid Search

36 TP/SL combinations: optimal zone search by profit factor and max drawdown

// services

Services

consultation

$100

~2 hours

+Analysis of your strategy
+Data-driven evaluation
+Optimization recommendations
+Risk & edge-case analysis

book()

backtest

$500

2–3 days

+Strategy formalization into a model
+Backtest on tick data
+Metrics: WR, PF, DD, Sharpe
+Lookahead bias verification
+Report with equity curve charts

run_backtest()

research

$2000

from 1 week

+Hypothesis discovery on BigData
+Testing N hypotheses
+Custom feature engineering
+Strategy with parameters
+Documentation: entry/exit/params

start_research()

// ready_strategies

Ready-Made Strategies

Backtested and verified strategies available for purchase. Equity curves and metrics included.

strategy_001

BTC Futures — Crossing Strategy

$1000

+ 20% of profit

Backtest period

1.5 years

Trades

Result

+45%

purchase_strategy()

// workflow

Workflow

Request

Task description: strategy, instrument, timeframe

Estimation

Scope estimation, timeline confirmation

Work

Research, testing, analysis

Result

Report, metrics, recommendations

Support

Follow-up questions on results

// faq

FAQ

Which markets does the system work with?

Infrastructure is tailored for Binance Futures (BTCUSDC). The methodology is applicable to any liquid instrument with access to tick data.

What data is used?

aggTrades (every trade), aggregated into minute metrics: buy/sell qty, divergences, normalized indicators, volatility, drift. Not OHLCV.

Profit guarantee?

No. The guarantee is research quality, backtest correctness, absence of lookahead bias, and an honest report. If the strategy is unprofitable — that will be in the report.

Can the found strategy be automated?

The strategy is delivered in formalized form: entry/exit formulas, parameters, conditions. Ready for implementation in any framework. Bot development is a separate service.

What if the strategy turns out to be non-working?

Report with metrics, equity curve and explanation of reasons. A negative result is still a result. Better to learn this from a backtest than from your balance.

Result format?

Consultation: Zoom + summary. Testing: report (equity curve, metrics, trade log, recommendations). Strategy research: documentation + formalized rules + parameters.

Strategies for algorithmic trading