Custom Trade Logging System for AI Trading Bots

Note: when a bot loses money, understanding the reason without detailed context is impossible. Logs with only price and volume don't provide answers. A complete chain is required: the feature vector at the moment of signal, model version, prediction (score/probability), execution conditions (slippage, commission, spread), and final P&L for each position. Without this, debugging is guessing on feature vectors. We design a trade recording system that records every trade with full context. With over 8 years in algorithmic trading and 50+ custom logging implementations, we deliver proven solutions. Our team has extensive experience in AI trading systems and numerous turnkey implementations.

The first thing teams encounter is the gap between signal and execution. The model predicts a movement, but the broker executed the order at a different price due to latency or slippage. The log must record the submission timestamp, reception timestamp, spread, and slippage amount. Otherwise, you won't understand where the profit was lost.

We use ClickHouse—a columnar DBMS that provides 50,000 trades/sec write and 5-10x data compression. This allows storing billions of records and performing ClickHouse analytics.

Critical importance of trade logging for AI trading

Without detailed recording, you cannot:

• recover the cause of a losing trade (model error, latency, market shock);
• provide regulatory reports on compliance with trading limits;
• improve the model: P&L attribution by factors (model version, symbol, feature set) requires clean historical data.

Comparison of logging approaches:

ClickHouse is 10x faster than PostgreSQL for log aggregation with volumes over 10 million records (ClickHouse Benchmark).

Essential data to log

These data allow fully reproducing the trade history.

How we implement the logging system

We use ClickHouse as the primary storage, Python for ETL, Airflow for orchestration. Below is a code snippet that writes signal, order, and fill in one transaction (via INSERT queries).

import json
import uuid
from datetime import datetime
from clickhouse_driver import Client

class TradingLogger:
    def __init__(self, clickhouse_host: str):
        self.ch = Client(clickhouse_host)
        self._ensure_tables()

    def log_signal(self, symbol: str, features: dict,
                   prediction: float, model_version: str) -> str:
        signal_id = str(uuid.uuid4())
        self.ch.execute(
            """INSERT INTO trading_signals VALUES""",
            [{
                'signal_id': signal_id,
                'timestamp': datetime.utcnow(),
                'symbol': symbol,
                'model_version': model_version,
                'prediction': prediction,
                'features': json.dumps(features),
            }]
        )
        return signal_id

    def log_order(self, signal_id: str, symbol: str, side: str,
                  qty: int, order_type: str, limit_price: float = None):
        self.ch.execute(
            """INSERT INTO orders VALUES""",
            [{
                'order_id': str(uuid.uuid4()),
                'signal_id': signal_id,
                'symbol': symbol, 'side': side, 'qty': qty,
                'order_type': order_type, 'limit_price': limit_price or 0,
                'submitted_at': datetime.utcnow()
            }]
        )

    def log_fill(self, order_id: str, fill_price: float,
                 fill_qty: int, commission: float):
        self.ch.execute(
            """INSERT INTO fills VALUES""",
            [{
                'fill_id': str(uuid.uuid4()),
                'order_id': order_id,
                'fill_price': fill_price,
                'fill_qty': fill_qty,
                'commission': commission,
                'filled_at': datetime.utcnow()
            }]
        )

Case study: one client saved 40% debugging time after implementing our system. Previously, finding an anomalous trade took 2–3 hours; after, 15 minutes. This became possible thanks to full context: feature vector, model version, slippage. Implementation cost starts from $5,000 for a single-bot setup, and clients typically see a 30% reduction in debugging costs. Our system enables real-time trade monitoring and alerting. We provide a 30-day satisfaction guarantee and all our engineers are ClickHouse certified.

Analyzing P&L by models

ClickHouse allows efficient analysis of millions of trades with P&L attribution:

-- P&L attribution by model and symbol
SELECT
    model_version,
    symbol,
    sum(realized_pnl) as total_pnl,
    count() as trade_count,
    avg(fill_price - requested_price) / avg(fill_price) * 10000 as avg_slippage_bps
FROM fills
JOIN orders USING order_id
JOIN trading_signals USING signal_id
WHERE filled_at >= today() - 30
GROUP BY model_version, symbol
ORDER BY total_pnl DESC;

Comparison of ClickHouse vs PostgreSQL for log analytics:

Data from ClickHouse Benchmark

Typical implementation mistakes

• Missing unique trade identifier—breaks signal-order-fill link.
• Logging only successful orders—you lose information about rejected/cancelled orders.
• Not saving feature vector—impossible to reproduce the model's decision post-factum.
• Storing logs in a single table without partitions—historical queries slow down by an order of magnitude.

Process

1. Analysis—study current architecture, define field list (features, orders, fills).
2. Design—develop ClickHouse schema with tags and partitions (day/model).
3. Implementation—write TradingLogger class, integrate with broker API. Include Grafana dashboard.
4. Test—load historical data, verify signal->order->fill links.
5. Deployment—CI/CD, alert monitoring (no logs for >5 minutes).

Timeline: 2 to 4 weeks depending on number of models and data sources.

What's included

• Source code of logging library with documentation in English.
• ClickHouse migration scripts (table creation, materialized views).
• Example Grafana dashboard with key metrics (P&L, slippage, trade count).
• Real-time trade monitoring and alerting.
• 2-hour training webinar for the team.
• 1 month of support after deployment.

Contact us for a consultation—we'll assess your project in one business day. Order the trade recording system implementation and gain full control over your AI trading bot trades. Get a consultation—our engineers will help configure logging for your strategy.

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.