How to evaluate trading platforms
How to evaluate trading platforms? Start with what a demo will never show you: what the system does on a bad day, when a price gaps, a feed dies, or an order does not come back. Remember, the polished walkthrough is built to impress.
Almost every platform looks solid from the outside. Clean screens, quick fills, a long list of logos. The catch is that the parts which decide whether you keep your money are the parts a sales walkthrough never reaches. Even the giants get caught: in January 2025 the SEC fined the quant firm Two Sigma $90 million after a flaw in its own models sat unfixed for about four years. If it can hide there, it can hide in the platform you are about to trust.
You do not need to be an engineer to tell a solid platform from a slick one. You need to know which questions cut through the pitch, and what a good answer sounds like. That is what this page gives you.
Start with the one thing a demo is built to keep you from seeing.
What you are paying for in a trading platform
The screen you are shown is the smallest part of a trading platform. Underneath sits the machinery that matters: the part that takes your order, sends it to a market, checks the fills come back, keeps the records, and controls who is allowed to do what. The interface is only the reception desk.
When that machinery is weak, you do not notice on a good day. You notice the first time the market moves fast: an order that fills at a worse price, a report that does not add up, a data feed that drops while the system keeps trading on old prices. A feature list will never show you this. The first bad week will.
No platform makes a strategy profitable. But a weak one quietly bleeds returns and makes a good strategy impossible to trust. The wider picture of everything a platform has to handle sits on our algorithmic trading platform hub.
Why a demo cannot tell you if a platform is any good
A demo is not a document you can study. It is a guided tour. Someone who knows the platform inside out drives, on a day they picked, with data they picked, taking only the routes they know work. It is a sales conversation, and it is very good at being one. What it cannot do is show you the platform on a day nobody rehearsed.
The things that cost you money live one level below the tour. How much of the workflow is genuinely wired together, and how much was stitched up the week before your call? What happens when a data feed dies mid-session? If one person holds the whole thing together in their head, what happens when they are on holiday?
Put plainly: most demos are designed to keep you away from the exact questions that decide whether the platform is any good.
| What the demo is built to show | What only live trading reveals |
|---|---|
| A clean order ticket and instant fills | How the system copes with a fill that never comes back the normal way |
| Integration logos and connected feeds | Which connections are looked after, and which were wired up the week before |
| A stable market-data panel | What the platform does when a feed dies mid-session |
| Role-based dashboards | The one person who understands the whole thing, and what happens when they are away |
| A polished onboarding flow | How many months your team needs before it can run this without the vendor on the phone |
| Green monitoring lights | Which alerts fire, to whom, and how fast, at 3pm on the worst day of the quarter |
A demo is a fine first look. Use it to see whether the platform is pleasant to operate and whether it fits how you work. What you cannot do is spot what it hides, because by definition you cannot see it. So you do not go looking. You ask. Here are the questions that surface what a tour skips, and where on this site we go deeper on each.
- Ask what happens when a data feed dies mid-session, and who finds out first. Good and bad answers look very different: our market data and connectivity page shows which is which.
- Ask how the platform handles an order that fills badly, or not at all. This is execution, and it is where most money quietly leaks, which is why we give slippage its own page.
- Ask who, on your side, could keep it running without the vendor on the phone. If the honest answer is nobody, that is the finding.
- Ask to see the platform put through the five-point check we run on any platform, the one in our scoring below.
The five places a good platform proves itself
Once you get past the tour, strong platforms pull away from weak ones in five places. You do not have to test them yourself. For most there is a public document or a law you can hold the provider to; the last one is about people, so you test it by asking. Either way it comes down to evidence rather than opinion.
Whether it handles your orders when it counts
The first thing to check is whether the platform can get your order to the market and back without losing money on the way. A good provider can walk you through exactly what happens from the moment an order is placed to the moment it is confirmed and recorded.
Here is why it matters. On 1 August 2012 the trading firm Knight Capital pushed a faulty software update and fired millions of unintended orders in 45 minutes, losing about $440 million, close to four times what it had earned the quarter before.
The safeguards that would have stopped it are the ones you are checking for, later written into law as the SEC Market Access Rule. If a provider is vague here, trust nothing built on top of it.
How well it connects to your brokers and data
A platform is only as good as its connections: to your broker, your data feeds, and everything else it has to talk to. Products quietly oversell here. Linking to one broker in a narrow way proves little about handling many, across venues, under real load.
Ask about the limits. Interactive Brokers’ Web API allows 10 requests a second and returns an error the moment you go over. A provider who cannot talk at that level of detail has not run at scale. Standards like the FIX protocol and FIXatdl exist so these connections behave the same way every time.
Connections are now a watched risk. FINRA’s 2026 oversight report, published December 2025, points to a rise in outages at firms’ outside vendors and expects firms to keep a list of every one they depend on. A platform that cannot name its own dependencies cannot help you manage yours. Our client integration guide goes deeper here.
Whether you can trace what happened
When something goes wrong, can you reconstruct exactly what happened, when, and who did it? Ask how the platform keeps its records straight. Interactive Brokers’ own developer documentation admits that duplicate status messages are normal, and that some fills only appear on a separate report and have to be matched by hand.
That is the messy reality of live trading. A platform that pretends it is clean has not handled enough volume to know better.
Whether it holds up on a bad day
The last place to look is what happens when things break: backups, monitoring, and how fast a human is told. Mature venues treat this as routine. Cboe Europe publishes an emergency playbook, refreshed in 2024, covering backups, recovery testing, and how quickly they tell people when something goes wrong.
Two laws now set a floor you can hold a provider to. MiFID II Article 17, in force since January 2018, requires trading systems to be resilient and stress-tested, with a continuity plan.
DORA, the EU’s operational-resilience law, has applied since 17 January 2025 and goes further: documented risk management, mandatory testing, incident reporting on the clock, and oversight of critical technology vendors. Ask to see the provider’s version of both, and how it fits the wider operational risk picture. A provider running real money already has the paperwork.
Whether your team can run it without the vendor
A platform only helps if the people who bought it can run it. There is a real gap between a smooth sales team and a system your own people can operate on a bad Tuesday afternoon.
Ask who on your side would own it, and speak to reference clients who have run it for years rather than weeks. If keeping it alive always needs the vendor on the phone, you are renting a dependency, and that bill only grows.
How to check all this without being an engineer
You are not going to audit the code, and you do not need to. Every check above comes down to requesting a few specific things and reading how the provider responds. The request itself is a test: a strong operation hands these over, a weak one stalls.
- Their pre-trade controls, in writing. The hard limits, the off-switch, and who owns it. If the answer is verbal or vague, treat that as the finding.
- A recent reconciliation report. Ask them to show how one day’s intended trades matched the fills that came back. If they cannot produce one, you have your answer.
- Their vendor and dependency list. The inventory FINRA now expects every firm to keep. It shows what your uptime rests on, across the trading infrastructure underneath.
- The result of their last disaster-recovery test, with a date. A written policy is not the same thing. A real test has a date, a scenario, and a list of what broke.
- A walk through the system’s architecture. How the pieces fit, and where the single points of failure sit. Our platform documentation guide shows how to read one.
Notice the pattern. You are not testing the software. You are testing whether the people behind it can show their work, on demand, without flinching.
The five questions that matter most, and they are all about execution
You can shorten any platform review to a handful of questions. Keep them on execution, because execution is where a strategy lives or dies. Everything else is easier to fix than a system that cannot get your trades done properly. If a provider answers these five well, they have earned a longer look.
- When an order fills at a worse price than expected, or does not fill at all, what does the system do, and where would I see it?
- How does the platform make sure every trade it thinks it made was in fact carried out and recorded correctly?
- If a data feed or the broker connection drops mid-session, what happens to my open orders?
- In a fast, busy market, does the platform keep up or fall behind, and how do you know?
- When something breaks at the worst possible moment, who gets alerted, and how quickly?
How we put this to work: The Review
This is not theory for us. We built The Review at Algotrader.ch for exactly this problem: an independent look at real algo and quant operations, starting with risk management, the strategies, the people who run them, and the platforms underneath.
We score every operation the same way, across five dimensions. A platform is judged mostly on two of them: how well it executes, and how well it runs day to day. Those two break into the five practical checks below. The first operation we scored shows it end to end.
Here is the checklist we use on a platform. It is yours to run on any provider you are weighing, no jargon required.
| What we check | Strong signal | Weak signal |
|---|---|---|
| Getting orders done | A documented path from order to confirmation, with limits and a named owner for the off-switch | “The platform handles it” |
| Connections | Looked-after, repeatable links across brokers and venues | Logos, and connectors wired up per client |
| Traceability | You can reconstruct who did what, when; records reconcile | Partial logs, gaps patched by hand |
| Holding up on a bad day | Recovery testing, incident drills, DORA and MiFID II-grade discipline on file | An uptime promise with nothing behind it |
| Running it in real life | Training built around your team, who become self-sufficient | Permanent dependence on the vendor |
Most platforms clear two of these five and hope the demo carries the rest. The gap between the two columns is where your money leaks. The full method is published in the open in our scoring framework.
Questions we get asked about trading platform due diligence
See which platforms clear the bar
The Review is our independent scorecard for real algo and quant operations, and the platforms running them, judged on the exact checks you just read. Coverage is selective and growing, so instead of sitting through another demo you can start from the algorithmic trading platforms that have already held up.