TacticTrades
STRATEGY LIBRARY / RESEARCH NOTES

Understand the idea.
Interrogate the edge.

A field guide to widely used algorithmic trading approaches—from trend following and market making to statistical and crypto arbitrage. Each method has a mechanism, a set of assumptions, and conditions in which it can fail.

Browse strategy families ↓Jump to arbitrage
08Core strategy familiesDirectional, relative value, execution
05Arbitrage approachesVenue, cycle, basis, statistical
06Execution checksFrom fee model to failure handling
FIRST PRINCIPLE

There is no best algorithm
without a specific problem.

Algorithms encode assumptions about how markets behave. Trend systems seek persistence; mean-reversion systems expect a return toward a reference; market makers manage inventory while quoting liquidity; arbitrage systems compare related prices; execution algorithms try to manage the cost of completing an order. These are different jobs, and their performance should not be compared as if they were interchangeable products.

The pages below are educational descriptions of established strategy families. They are not claims that TacticTrades currently deploys every method, nor are they live signals, investment advice, or evidence of profitability.

01 — STRATEGY FAMILIES

Know the mechanism.
Know the failure mode.

Use each entry as a research starting point. Implementation details, market access, costs, and controls determine whether a theoretical method is practical.

01 / DIRECTIONAL↗

Trend following & time-series momentum

Prices that have moved persistently in one direction may continue to do so over a chosen horizon. Systems typically define a lookback, a trend measure, and a rule for exiting when the signal weakens.

COMMON TOOLS

Moving-average crossovers, breakout levels, time-series momentum, volatility targeting, trailing exits.

WHAT CAN GO WRONG

Choppy markets can create repeated whipsaws. Trends reverse, signals arrive late, and volatility scaling cannot eliminate gap risk.

VALIDATION QUESTIONS

Test multiple regimes, delayed execution, turnover, sensitivity to lookback choices, and the distribution of drawdowns—not just total return.

02 / RELATIVE VALUE↗

Mean reversion & pairs trading

A price, spread, or portfolio may move away from a defensible equilibrium and later move back. Pairs trading expresses this through relative performance rather than relying only on the broad market direction.

COMMON TOOLS

Spread construction, rolling hedge ratios, z-scores, stationarity checks, cointegration analysis, half-life estimates.

WHAT CAN GO WRONG

A relationship can break permanently. Correlation alone does not prove cointegration, and a spread can keep widening longer than capital or margin allows.

VALIDATION QUESTIONS

Use point-in-time data, out-of-sample periods, realistic hedge ratios, borrow/funding costs, and explicit exits for structural breaks.

03 / QUANTITATIVE↗

Statistical arbitrage

A family of quantitative relative-value methods that look for temporary pricing relationships across securities, baskets, or related instruments. The term describes a research approach, not a guaranteed riskless profit.

COMMON TOOLS

Cross-sectional ranking, factor neutralisation, residual models, cointegration, portfolio construction, turnover and exposure constraints.

WHAT CAN GO WRONG

Model error, crowded trades, hidden factor exposures, unstable relationships, and transaction costs can overwhelm small statistical edges.

VALIDATION QUESTIONS

Measure factor and sector exposure, use robust out-of-sample tests, account for multiple testing, and monitor decay between predicted and realized performance.

04 / LIQUIDITY PROVISION↗

Market making & spread capture

A market maker posts bids and offers, seeking to earn spread while managing inventory and the risk that incoming orders are better informed than its quotes.

COMMON TOOLS

Fair-value estimates, quote skew, inventory limits, queue position, adverse-selection measures, cancel/replace controls.

WHAT CAN GO WRONG

A displayed spread is not guaranteed income. Toxic flow, rapid repricing, inventory imbalance, fees, and queue priority can turn spread capture negative.

VALIDATION QUESTIONS

Model fills rather than assuming them, include maker/taker fees, measure markouts after fills, and stress-test inventory during sharp moves.

05 / CROSS-SECTIONAL↗

Relative momentum & ranking

Rank a universe by recent performance or another signal, then allocate according to a defined selection and rebalancing rule. The result depends on the universe and the portfolio construction as much as the ranking formula.

COMMON TOOLS

Momentum scores, cross-sectional ranks, rebalance schedules, turnover controls, volatility adjustment.

WHAT CAN GO WRONG

Crowding, reversals, universe survivorship, stale constituents, and rebalance costs can distort historical results.

VALIDATION QUESTIONS

Use point-in-time constituents, include delisted assets where relevant, and report turnover, concentration, and performance across regimes.

06 / EXECUTION↗

VWAP, TWAP & participation algorithms

Execution algorithms schedule a parent order over time or relative to market volume to manage implementation costs. They are execution methods, not standalone predictions that a market will move profitably.

COMMON TOOLS

Volume-weighted average price (VWAP), time-weighted average price (TWAP), participation-of-volume, arrival-price benchmarks.

WHAT CAN GO WRONG

Unexpected volume, market impact, urgency, volatility, and thin liquidity can make a schedule inappropriate. Matching a benchmark does not itself mean the trade was profitable.

VALIDATION QUESTIONS

Compare implementation shortfall and slippage to a suitable benchmark, with order size, urgency, and market conditions segmented.

07 / PRICE STRUCTURE↗

Breakout & volatility expansion

A system looks for price leaving a defined range or volatility compression resolving into a larger move. Entry and invalidation rules must be specified before the breakout is observed.

COMMON TOOLS

Range highs/lows, channel breakouts, ATR filters, volatility compression, volume and liquidity filters.

WHAT CAN GO WRONG

False breakouts, gaps, thin books, and delayed entries can make the risk/reward profile very different from a chart illustration.

VALIDATION QUESTIONS

Include slippage on gaps, define stops and sizing before entry, and avoid selecting thresholds solely because they fit one sample.

08 / RANGE-BASED↗

Grid strategies

Place orders at defined price intervals around a reference range, attempting to trade oscillations. Grid logic changes exposure as price moves and can accumulate inventory during a sustained trend.

COMMON TOOLS

Grid spacing, inventory caps, range resets, volatility-aware spacing, capital allocation.

WHAT CAN GO WRONG

A grid is not inherently market-neutral. Trending prices, fees, funding, and insufficient capital can leave the system holding a deteriorating position.

VALIDATION QUESTIONS

Stress one-way markets, quantify worst-case inventory, account for every order and fee, and specify when the grid is disabled or re-centred.

02 — CRYPTO & RELATIVE-VALUE ARBITRAGE

Price differences are easy
to see. Hard to capture.

An arbitrage calculation is only as good as its fee schedule, executable depth, timestamps, order handling, and treatment of incomplete execution.

01
VENUE A ↔ VENUE B

Cross-exchange spot arbitrage

Compare executable buy and sell prices across venues for the same asset. The real calculation must include order-book depth, fees on both venues, inventory location, transfer restrictions, withdrawal costs, and the possibility that one side fills while the other does not.

THE CATCH

A price gap can disappear before execution. Pre-funded inventory reduces transfer delay but introduces capital allocation, custody, venue, and rebalancing risk.

02
ASSET A → B → C → A

Triangular arbitrage

Evaluate a closed conversion cycle across three trading pairs. Each conversion changes the quantity available for the next leg; bid/ask direction, precision, minimum order sizes, fees, and depth must be applied sequentially.

THE CATCH

A theoretical cycle can become negative after rounding and fees. One rejected or partial order can leave an unhedged asset exposure rather than a completed cycle.

03
MODELLED RELATIONSHIP

Statistical arbitrage

Estimate a relative-value relationship between instruments and look for deviations that are large enough to justify trading after costs. The opportunity is statistical and model-dependent, not a risk-free price lock.

THE CATCH

The relationship may change regime, the hedge may drift, and convergence timing can exceed risk limits.

04
SPOT ↔ DERIVATIVE

Cash-and-carry & funding basis

Compare spot and derivative prices while accounting for maturity, funding payments, financing, margin, collateral, borrow availability, and the cost of maintaining both legs.

THE CATCH

Funding rates change, basis can widen before expiry, liquidation rules differ, and hedges can fail under stressed liquidity.

05
DEPTH / VWAP / SLIPPAGE

Order-book & liquidity dislocations

Estimate whether a displayed price difference survives execution at the intended notional. Weighted average price and depth-by-level analysis help distinguish a top-of-book headline from an executable trade estimate.

THE CATCH

Books can be stale, spoofed, rapidly consumed, or inconsistent across timestamps. The estimated fill price is still not a fill guarantee.

A SIMPLE NET-EDGE MODEL
Expected net edge = executable proceeds − acquisition cost − all costs

“All costs” can include trading fees, spread, slippage, funding, borrow, transfer or rebalancing costs, and a buffer for latency and incomplete fills. If those inputs are stale or uncertain, the result is an estimate—not proof that the trade is profitable.

03 — BEFORE A SIGNAL BECOMES AN ORDER

Execution is part
of the strategy.

In fast or thin markets, small modelling errors can erase a narrow edge. These checks should be explicit in research and production design.

01

Fees

Use the current account-specific maker/taker tier and the actual fee currency for each leg.

02

Executable size

Calculate the expected average fill from available depth at the target notional, not only best bid and ask.

03

Latency & freshness

Reject stale or out-of-order books; measure the age of both sides and the time from signal to order.

04

Partial fills & leg risk

Define what happens if one leg fills, the other rejects, or available liquidity vanishes.

05

Capital & constraints

Check balances, precision, minimum notional, rate limits, margin, borrow, and venue-specific order rules.

06

Net edge & stop conditions

Require a positive, adequately buffered net estimate and stop when realized results contradict the model.

04 — EVALUATION & EVIDENCE

A beautiful equity curve
can still be misleading.

Backtests can be distorted by look-ahead bias, survivorship bias, data snooping, parameter overfitting, unrealistic fill assumptions, and omitted costs. A robust review separates research from validation, keeps a final holdout untouched, and reports more than a headline return.

Useful measures include maximum drawdown, volatility, turnover, tail losses, trade count, exposure, capacity, sensitivity to parameters, and the difference between predicted and realized execution. None guarantees future performance, but each can expose a weakness hidden by a single metric.

01Data integrityTimestamp alignment · missing data · point-in-time universe
02Realistic costsFees · spread · slippage · funding · market impact
03Out-of-sample evidenceWalk-forward tests · holdout data · regime diversity
04Operational stressLatency · partial fills · rejected orders · stale books
05Ongoing monitoringDrawdown limits · model error · stop conditions
FURTHER READING

Continue the research.

These independent resources provide deeper background on mean reversion, pairs trading, statistical tests, and execution algorithms. They are educational references, not endorsements of a particular strategy.

QUANTSTARTBasics of Statistical Mean Reversion TestingRead article ↗QUANTSTARTBacktesting an Intraday Mean Reversion Pairs StrategyRead article ↗INTERACTIVE BROKERSTrader Workstation guide: VWAP & TWAP order typesOpen guide ↗
IMPORTANT RISK NOTICE

Understanding a method
does not make it safe.

All trading involves risk, including loss of capital. Arbitrage can carry execution, liquidity, counterparty, custody, transfer, and technology risks. Historical or simulated results do not guarantee future outcomes. This material is general education, not personalised financial advice or a recommendation to trade.

Read the full terms ↗