PredictEngine review
Our verdict
The most capable-sounding product here on features — cheap, multi-venue, no-code, with a free simulation tier that is a legitimately good way to learn. It ranks last on our index anyway, because a hosted service that runs trades for you and does not state whether it holds your keys has failed the one disclosure that matters most. The unverified 5–20% monthly figure does not help. Clarify custody and this would move up sharply.
How it works
A no-code platform for building and deploying prediction-market bots. You describe a strategy, and the platform watches the order book and fires orders when triggers hit, running on its own cloud infrastructure so your machine can be off. It includes a simulation mode for paper trading against live data, built-in risk caps for daily loss, position limits and concurrent exposure, order-book depth scanning to limit slippage, and support for Polymarket, Kalshi and Limitless.
Strengths and weaknesses
What works
- Cheapest paid tier in this review at $19/month, with no per-trade fee and no profit share.
- Free simulation tier lets you build and test a strategy at zero cost and zero risk.
- Multi-venue support across Polymarket, Kalshi and Limitless is unique here.
- Always-on hosted execution removes the uptime burden of self-hosting.
- No-code builder genuinely lowers the barrier for non-programmers.
What doesn't
- Custody model is not stated anywhere we could find. For a hosted service that executes trades on your behalf, this is the most important disclosure there is, and its absence is why this ranks last.
- The claim that profitable strategies typically return 5–20% per month is presented without verification, sample size or methodology.
- No published performance data or user count.
- Hosted execution means the operator's infrastructure is a dependency you cannot audit or replace.