The Review
Most directories in the algo & quant space publish a leaderboard and call it analysis. Some add a Sharpe ratio column. A few include a short strategy description written by the manager. None of those things answer the question a serious investor needs answered before putting capital to work.
The Review is built around a different set of questions. How was this strategy built, and what did the developer throw away before landing on the current version. How does the risk framework behave when it gets hit, not when markets cooperate. What happens if the developer cannot reach a keyboard for five days. Five dimensions. Twenty-five criteria. The results are on the listing page. Behind all of it sits one goal: strategies steady enough to compound a portfolio.
What The Review is and what it is not
The Review is an editorial assessment. The algotrader.ch team evaluates each covered strategy across five dimensions using the same criteria for every strategy. Evidence is the basis: DDQ responses, live fill data, code deployment logs, and a direct technical walkthrough with the developer. The assessment is not a profile the developer writes about themselves and submits for publication. That is a different product and not what this is.
- Not a ranking. Strategies are assessed against a fixed framework, not against each other. The bars on a listing show where a strategy sits on each criterion. They do not say which strategy is better than which other. That question depends on the investor.
- Not paid placement. No covered strategy has paid for inclusion or for a particular outcome in the assessment. That will not change.
- Not investment advice. An assessment tells you what we found. What you do with that information is your decision.
Coverage is selective. The editorial team decides what to review based on what is interesting, credible, and useful for the audience. Most enquiries do not result in a full assessment. That selectivity is the point.
The five-dimension framework
Every strategy is assessed across the same five dimensions. Within each dimension, five specific sub-criteria are scored. The dimensions run in order of importance. Risk management is the first question, not the fifth.
Will this strategy survive a bad month? Not in a controlled test. In the conditions that actually showed up in the last three years: the October 2023 gold spike, the March 2024 equity selloff, the 2022 rate shock. We look at track record length, how the strategy behaved across those periods, how position sizing is calculated, and whether the developer can name what would break the edge without reaching for vague language about tail risk.
- Length and quality of the live track record
- Multi-environment evidence, not just the recent good years
- Position-sizing methodology: logic, consistency, documentation
- Stress event handling: specific dates, specific numbers
- Failure mode awareness: can the developer name what kills the edge?
When conditions turn bad, what does the system actually do? This is a different question from risk management. Risk management is the strategic layer. Protection controls are the tactical one. A circuit breaker that fires on a 2% intraday gold spike. A spread filter that cancels orders when liquidity is thin. We look for evidence that these rules exist in the code, not just in the documentation, and that something other than a calm market has tested them.
- Slippage controls: tolerance thresholds and abort logic
- Volatility-spike response: size reduction, pause, or pass-through
- Liquidity safeguards: session blacklists, spread filters, volume thresholds
- Stale-position controls: time-based stops enforced by the system
- Circuit breaker: pre-committed triggers, has it fired?
Paper alpha dies in execution. A strategy that looks clean in a backtest can be taken apart by slippage in thin sessions, by orders that fill badly during news events, by a system that assumes fills happen cleanly when they routinely do not. We look at whether backtest slippage assumptions hold up against actual fills, how the system handles partial fills and rejected orders, and whether anyone reconciles intended trades against actual trades at the end of each session.
- Slippage assumptions in the backtest: realistic or flat?
- Partial-fill and no-fill handling: documented behaviour
- Rejected-order handling: classification, retry logic, alerting
- Trade reconciliation: intended vs actual fills, end-of-session
- Execution window discipline: avoids predictably bad session times
Most operational failures in retail systematic trading are not strategy failures. They are deployment accidents. A code change that went to the live server without testing. A data feed that dropped while the system kept trading on stale prices. A developer who got sick and no one else knew the login. We look at who controls the stack, how code changes are tested before going live, what happens when the primary broker goes down, and whether anyone else could run the strategy if the developer is not available.
- Stack control: does the developer own the failure points?
- Deployment: staged environment, version control, rollback
- Unattended monitoring: heartbeat, deviation, anomaly alerts
- Vendor failover: secondary feeds, graceful degradation
- Continuity: can someone else run this if the developer is unavailable?
How did the strategy get here, and has the developer stayed honest about what the research actually showed? A graveyard of discarded strategies is more informative than a clean backtest. We look at what was tested and thrown away, whether the out-of-sample period was genuinely untouched during development, and how closely live returns track the backtest. A 10 to 15% Sharpe degradation from backtest to live is normal. A perfect match is a flag.
- Research process: documented graveyard of discarded strategies
- Out-of-sample validation: was the held-out period untouched?
- Live vs backtest: gap, direction, explanation
- Parameter stability: does nudging the inputs collapse performance?
- Change management: infrequent, dated, not triggered by recent losses
How scoring works
Each of the 25 sub-criteria gets a score from 1 to 5. The score maps to a bar on the listing page. An empty bar means the evidence is weak or missing. A full bar means it is among the best we have seen in the category. Two things are worth saying clearly about how that works.
| Score | What it means |
|---|---|
| 1 | Concerning. Evidence is missing, contradictory, or actively bad. A real risk to the investor. |
| 2 | Below standard. The developer is aware of the question but the answer is weak. |
| 3 | Standard. What you would expect from a competent professional setup. The base case. |
| 4 | Strong. Visibly above peers. Specific, evidenced, well-documented. |
| 5 | Best in class. Among the best we have seen across the category. Rare. |
Scores are assigned by the editorial team based on what was reviewed. The developer does not control the score. Their responses are the primary input, but those responses are weighed against the evidence, not taken at face value.
A strategy that scores 4 on risk and 1 on operations is telling you something. The bars make that pattern visible without requiring you to read 25 data points first.
There is no overall score across dimensions. An allocator who weights risk management above everything else will read the five bars differently from one who focuses on research discipline first. The framework does not impose a weighting. Its job is to make the evidence legible. The decision is yours.
Current coverage
The Review is selective. Coverage starts small and grows as assessments are completed. A listing represents a full assessment by the editorial team, not a submitted profile that has been lightly checked. Speed is not the objective.
Further strategies are currently under assessment. Listings are published when the work is done.
Strategies that compound a portfolio
The framework has one outcome in mind: a strategy an investor can leave running for years. Compounding is the reason. A steady monthly return, retained and never broken by an oversized loss, is what turns a trading account into wealth. The arithmetic sits on our compounding portfolio growth page, and it is demanding: 3% a month with minimal losses compounds a portfolio to roughly 10× within seven years. Few strategies can hold a line like that for long. Finding the ones that can is what the five dimensions select for, and it is why capital preservation is scored before performance.
Read the first covered strategy through that lens. An average month of +2.81% next to a worst drawdown of -2.87% says less about the return than about the ratio between the two. A strategy that keeps its worst loss near its average month is structurally built for retention. What a return path like that becomes over one, five, or ten years is a question the compounding growth calculator answers in seconds, including the drawdown scenario most projections skip.
Getting a strategy reviewed
The Review does not run an open submission process. Coverage is editorial. The team decides what to assess based on what is credible, interesting, and relevant to the audience. If you run a systematic strategy and think it would pass a serious assessment, the right first step is a conversation, not a form.
- A live track record of at least 18 months. Real capital, real fills, real slippage. Paper trading records and backtests are not a substitute. Eighteen months is not a high bar, but it filters a meaningful share of what comes in.
- Evidence across all five dimensions. Performance data alone is not enough. Research notebooks, deployment logs, risk rule documentation. Something real in each area.
- A plain-English explanation of the edge. Two sentences. If you cannot say what the strategy does and why it makes money in two sentences, the assessment will reflect that.
- Willingness to name the gaps. Every strategy has them. The assessments that read best are the ones where the developer named the weak points before we found them ourselves.
What is white label trading?
White label trading lets a third party license an algorithmic strategy and offer it under its own brand, sharing the economics with the builder. Strategies carrying the white-label flag in The Review at Algotrader.ch passed our assessment, which puts capital protection before returns, and are worth considering as a white label trading strategy.
An assessment does more than inform investors. A strategy that has passed it carries the exact evidence a trading desk, a family office, or an asset manager asks for before licensing a system: risk rules that live in the code, execution reconciled against real fills, operations that survive the developer’s absence. Investors read the flag the same way: a strategy already running under more than one name says something about operational maturity.
And if you have an audience that trusts you but no algorithm of your own, the flag marks your entry point into managing. The numbers behind that role, from fee mechanics through capacity, are worked through under scaling a trading strategy. If you run capital and want a systematic book without building one, or you run a strategy and want it assessed and offered, get in touch.
For investors
If you are looking at a specific strategy and want to think through what to verify and where to push, a short conversation with the editorial team costs nothing. Not financial advice. A conversation with someone who has worked through the same assessment framework and can say what the bars on a listing page actually represent in practice.
Not investment advice. A serious conversation about due diligence.
Most conversations run 20 to 30 minutes. We share what we found beyond what is published on the listing, and what we think is worth verifying directly with the developer. We can make introductions, but only when the fit is genuine. Where one leads to an allocation, the manager may pay us a fee. It never affects a score.
The Review is an editorial product of algotrader.ch. Coverage decisions are independent. Nothing on this page constitutes investment advice or a recommendation to allocate capital to any strategy. Availability of any strategy covered is set by its manager and varies by jurisdiction.