Paper Trading System Development for AI Strategy Testing

Paper trading is a simulation of real trading without using real capital. Unlike backtesting (testing on historical data), paper trading runs in real time: the strategy receives live exchange data, generates signals, and sends virtual orders to a broker simulator. This reveals issues invisible in backtesting: data latency, order execution problems, technical failures. We build such systems turnkey so that your AI algorithm passes all validation stages without risking capital.

Why Paper Trading Is a Must

Even a perfectly tuned backtest can give false confidence. Look-ahead bias, overfitting on history, underestimated commissions—typical pitfalls. Paper trading in real time removes these risks: your AI model sees only the data that would have been available at decision time. The divergence in metrics between paper trading and backtest should not exceed 15% in Sharpe ratio—if larger, the strategy needs refinement.

According to the Wikipedia definition, paper trading is a simulation of trading, but in practice it gives much more: mistakes caught at the paper stage can cost up to $15,000 in real trading—that's how much our clients save on average. Paper trading is 3 times more accurate at uncovering execution problems than backtesting due to working with real latency and slippage.

System Architecture

[Market Data Feed] → [Data Normalizer] → [Feature Pipeline]
                                              ↓
                                      [AI Model Inference]
                                              ↓
                                      [Signal Generator]
                                              ↓
                                    [Risk Management Layer]
                                              ↓
                                    [Paper Broker Simulator]
                                              ↓
                              [Portfolio State] → [P&L Calculator]
                                              ↓
                                    [Monitoring Dashboard]

Components

Market Data Integration:

import asyncio
import websockets
import json
from dataclasses import dataclass

@dataclass
class MarketTick:
    symbol: str
    timestamp: float
    bid: float
    ask: float
    last: float
    volume: float

class AlpacaMarketDataFeed:
    def __init__(self, api_key: str, secret_key: str):
        self.ws_url = "wss://stream.data.alpaca.markets/v2/sip"
        self.headers = {
            "APCA-API-KEY-ID": api_key,
            "APCA-API-SECRET-KEY": secret_key
        }

    async def stream_quotes(self, symbols: list, callback):
        async with websockets.connect(self.ws_url, extra_headers=self.headers) as ws:
            # Subscribe to quotes
            await ws.send(json.dumps({
                "action": "subscribe",
                "quotes": symbols
            }))

            async for message in ws:
                data = json.loads(message)
                for item in data:
                    if item['T'] == 'q':  # quote
                        tick = MarketTick(
                            symbol=item['S'],
                            timestamp=item['t'],
                            bid=item['bp'],
                            ask=item['ap'],
                            last=(item['bp'] + item['ap']) / 2,
                            volume=item.get('bs', 0)
                        )
                        await callback(tick)

Paper Broker Simulator:

class PaperBroker:
    def __init__(self, initial_capital: float = 100_000):
        self.cash = initial_capital
        self.positions = {}  # symbol -> quantity
        self.orders = []
        self.fill_probability = 0.95  # 95% orders are filled

    def submit_order(self, symbol: str, qty: int, side: str,
                     order_type: str = 'market', limit_price: float = None):
        order_id = str(uuid.uuid4())
        order = {
            'id': order_id, 'symbol': symbol, 'qty': qty,
            'side': side, 'type': order_type,
            'limit_price': limit_price, 'status': 'pending',
            'submitted_at': datetime.utcnow()
        }
        self.orders.append(order)
        return order_id

    def process_tick(self, tick: MarketTick):
        for order in self.orders:
            if order['status'] != 'pending':
                continue
            if order['symbol'] != tick.symbol:
                continue

            # Simulate fill
            if order['type'] == 'market':
                fill_price = tick.ask if order['side'] == 'buy' else tick.bid
                self._fill_order(order, fill_price, tick.timestamp)
            elif order['type'] == 'limit':
                if (order['side'] == 'buy' and tick.ask <= order['limit_price']):
                    self._fill_order(order, order['limit_price'], tick.timestamp)
                elif (order['side'] == 'sell' and tick.bid >= order['limit_price']):
                    self._fill_order(order, order['limit_price'], tick.timestamp)

    def _fill_order(self, order: dict, price: float, timestamp):
        commission = price * order['qty'] * 0.0001
        if order['side'] == 'buy':
            cost = price * order['qty'] + commission
            if self.cash >= cost:
                self.cash -= cost
                self.positions[order['symbol']] = \
                    self.positions.get(order['symbol'], 0) + order['qty']
                order['status'] = 'filled'
                order['fill_price'] = price
        elif order['side'] == 'sell':
            if self.positions.get(order['symbol'], 0) >= order['qty']:
                self.cash += price * order['qty'] - commission
                self.positions[order['symbol']] -= order['qty']
                order['status'] = 'filled'

How the Broker Simulator Works

The broker simulator is a key component. It mimics order execution with a realistic fill probability (95% in our example) and commissions of 0.01%. This allows you to evaluate real slippage and spread impact. In live trading, latency-sensitive strategies may face worse execution—paper mode helps adjust the algorithm before going to market.

Real-Time Monitoring

The dashboard shows: realized and unrealized P&L in real time, open positions, trade log, equity curve, drawdown, benchmark comparison.

Key Metrics: Paper Trading vs Backtest

If paper trading results are significantly worse than backtesting, it indicates: overfitting to historical data, look-ahead bias in backtest, or underestimated transaction costs. Goal: deviation in Sharpe ratio < 15%.

Development Stages

What's Included

• Integration with exchange APIs (Alpaca, Interactive Brokers, Binance—your choice)
• Data normalization and feature pipeline module
• Interface for AI model (PyTorch/TensorFlow/JAX) with inference via vLLM or Triton
• Broker simulator with realistic fill probability and commissions
• Risk management: stop-loss, take-profit, position limits
• Real-time metrics dashboard (React + WebSockets)
• Documentation, team training, and 2 months post-launch support

How to Spot a Quality System

Request a demo run on 1-2 weeks of live data. A quality system will show stable p99 latency below 50 ms, 99% uptime, and correct order execution under high volatility. We guarantee these parameters—over 5 years we have delivered 30+ paper trading projects for funds and prop trading firms.

Timeline and Cost

Turnkey system development takes 30 to 60 business days depending on strategy complexity and number of instruments. Cost is calculated individually—contact us and we will estimate your project within 2 days. Get a no-obligation consultation. Contact us to discuss the details of your project.

MLOps: Infrastructure for Training, Deploying, and Monitoring ML Models

The model is trained, metrics — F1 0.94 on validation. Three months later in production, quality drops by 12%. No one knows when — there is no monitoring. It's impossible to retrain quickly — the training script is in a Jupyter notebook of a data scientist who has already left. Data for retraining is collected manually from three disparate systems. About half of the projects come to us with this pain. We build a turnkey MLOps platform: from experiment tracking to automatic deployment and data drift monitoring. We will assess your infrastructure in 1–2 weeks, and in 4–6 weeks you will get a basic MLOps core running in production. Our team has 10+ years of experience in ML infrastructure, over 50 implementations.

How does MLOps infrastructure benefit your ML projects?

Experiment Tracking and Reproducibility

Without tracking, an ML project turns into chaos: it's unclear which checkpoint is better, which hyperparameters were used, which dataset. Reproducing a result a month later is a quest.

Why is experiment tracking the foundation of reproducibility?

MLflow is an open source standard for tracking. It logs parameters, metrics, artifacts (models, graphs), and code. MLflow Model Registry is a centralized model storage with versioning and lifecycle stages (Staging → Production → Archived). Deployment via MLflow Serving or integration with external systems.

Typical initialization in code:

import mlflow

mlflow.set_experiment("fraud-detection-v2")
with mlflow.start_run():
    mlflow.log_params({"learning_rate": 3e-4, "batch_size": 64, "epochs": 10})
    mlflow.log_metric("val_f1", val_f1, step=epoch)
    mlflow.pytorch.log_model(model, "model")

This is the minimum. In production, we add logging of system metrics (GPU utilization, memory), dataset (hash, version), code (git commit hash). Weights & Biases — richer UI, collaboration features, sweep for hyperparameter optimization. MLflow — for on-premise deployment without external dependencies.

DVC (Data Version Control) — versioning of data and models on top of git. Data is stored in S3/GCS/Azure Blob, only metadata (hashes) in git. dvc repro reproduces the entire pipeline from raw data to metrics.

To ensure reproducibility of training, fix random seeds (torch.manual_seed, numpy.random.seed, random.seed) and record them in experiment metadata. Without this, debugging irregular results is painful. Log the dataset version (DVC hash) and git commit — then any experiment can be reproduced down to the byte.

Pipeline Orchestration: Kubeflow, Airflow, Prefect

A pipeline orchestrator becomes necessary when: A 100-line training script in cron is fine for simple tasks. But as soon as you have a multi-step pipeline (data loading → preprocessing → feature engineering → training → validation → deployment if quality above threshold), you need an orchestrator with retry logic, visualization, and alerts.

Kubeflow — Kubernetes-native orchestrator for ML (see Kubeflow). Each step is a Docker container. Supports parallel steps, conditional branches, artifacts between steps. Integrates with Katib (AutoML), KServe (serving), Feast (feature store).

Apache Airflow — more general DAG orchestrator. Wide ecosystem of operators (S3, Spark, DBT, Kubernetes). Easier to deploy if Airflow already exists in the company.

Prefect / Metaflow — less boilerplate. Prefect 2.x with @flow and @task decorators — quick start for small teams.

Typical training pipeline architecture on Kubeflow:

1. Data ingestion component — fetches data from S3/DB, validates schema via Great Expectations
2. Preprocessing component — transformations, normalization, train/val/test split
3. Training component — training on GPU, logging to MLflow
4. Evaluation component — metric calculation, comparison with baseline in Model Registry
5. Conditional deployment — deploy only if new model is better than current by >2% F1

Each component is a separate Docker image. Pipeline is versioned in git. Scheduled run (retraining once a week on new data) or manual.

Model Registry and Lifecycle Management

Model Registry is not just a checkpoint store. It is a centralized system that knows:

• Which model is currently in production (and with what metrics)
• History of all versions with training parameters
• Metadata: dataset, git commit, validation results
• Lifecycle stage: None → Staging → Production → Archived

MLflow Model Registry — standard. For enterprise — Vertex AI Model Registry (GCP), SageMaker Model Registry (AWS), Azure ML Model Registry.

Model promotion through stages: automatically move model to Staging after successful eval, then manual or automatic (during A/B test) promotion to Production. Rollback — switch to previous Production version in seconds.

Serving: From FastAPI to Triton Inference Server

Simple case. FastAPI + PyTorch/ONNX on one server — 80% of production ML deployments are exactly that. Sufficient for most tasks with load up to 100 req/s.

from fastapi import FastAPI
import onnxruntime as ort

app = FastAPI()
session = ort.InferenceSession("model.onnx", providers=["CUDAExecutionProvider"])

@app.post("/predict")
async def predict(request: PredictRequest):
    inputs = preprocess(request.text)
    outputs = session.run(None, {"input_ids": inputs})
    return {"label": postprocess(outputs)}

Triton Inference Server — production standard for high loads (500+ req/s). Dynamic batching, concurrent model execution, model ensemble. Supports TensorRT, ONNX, PyTorch TorchScript, TensorFlow SavedModel.

KServe — Kubernetes-native ML serving with autoscaling, canary deployments, A/B testing out of the box. Scale-to-zero for inactive models — savings on infrastructure up to 40% annually for a project with 10 models.

Monitoring: Data Drift, Model Drift, Infrastructure Metrics

Monitoring — what is usually done last and regretted first. Three levels.

Infrastructure monitoring. Latency (P50/P95/P99), throughput (req/s), error rate (4xx, 5xx), GPU/CPU utilization. Prometheus + Grafana — standard. Alert when P99 latency > threshold or error rate > 1%.

Data drift monitoring. Distribution of input data changes over time. Detect via PSI (Population Stability Index) for numerical features: PSI > 0.2 — strong drift. Chi-squared test for categorical, Kolmogorov-Smirnov test for continuous. Evidently AI — open source library with ready-made drift tests.

Model drift monitoring. If ground truth is delayed (e.g., we know conversion after a week) — monitor real metrics. If not — surrogate metrics: distribution of prediction scores, proportion of confident predictions.

Alerting. Three levels: INFO (minor drift, log it), WARNING (significant, notify team), CRITICAL (quality dropped below threshold — automatic switch to fallback model).

Why is data drift monitoring important?

Without it, you learn about model degradation only from user complaints or ringing SLA. A drift alert allows you to retrain the model in advance, before errors start causing losses. In one of our projects, PSI monitoring detected drift 2 days after a data source change — this saved the campaign.

Feature Store

Feature Store solves the training-serving skew problem. If preprocessing during training and inference is implemented in two different places — divergence is inevitable.

A Feature Store is needed when:

• Several models use the same features
• Features are computed from streaming data (real-time)
• Large team with different people on feature engineering and model training

Feast — open source Feature Store. Offline store (S3 + Parquet) for training, online store (Redis, DynamoDB) for low-latency inference. Feature definitions as code, materialization job syncs offline → online.

Tecton (commercial), Vertex AI Feature Store (GCP), SageMaker Feature Store (AWS) — managed options with less ops overhead.

CI/CD for ML

ML CI/CD is regular CI/CD plus specific ML steps.

ML-specific checks in CI:

• Reproducibility check: run training with a fixed seed, result must match
• Data validation: Great Expectations or Pandera on schema/distribution checks
• Model performance check: automatic eval on holdout, block merge if degradation > threshold
• Latency regression test: inference must meet SLA

GitOps for deployment. Merge to main → CI triggers training → eval → if passes → automatic deployment to Staging → smoke tests → manual promotion to Production or automatic upon successful canary.

Tools: GitHub Actions / GitLab CI for CI, ArgoCD for GitOps deployment on Kubernetes.

What's Included in MLOps Platform Development

We provide a full cycle of work, documentation, and team training.

Total time from audit to full MLOps platform: 3–5 months. Also possible phased launch: basic level (tracking + serving) in 4–6 weeks.

Cost is calculated individually based on data volume, number of models, and infrastructure requirements. Order an MLOps infrastructure audit — get a roadmap in 1–2 weeks. Contact us for a project assessment — we will send a preliminary estimate within 2 business days.

Note: warranty on architectural solutions — 12 months. We provide integration certificates with major cloud providers (AWS, GCP, Azure). During our work, we have not lost a single client after the first implementation — the experience of 50+ successful MLOps projects speaks for itself. Get a consultation on building an MLOps platform today.