The market for trading bots has been the same shape for a decade. Deposit funds into the bot's account. Pick a strategy from a dropdown, or pay a subscription for the "AI" version. Hope the bot's still around in six months and the founders haven't moved on.
Most of it hasn't worked. The opaque ones aren't smart; the smart ones aren't transparent. The ones that hold your funds keep failing — sometimes at the trade layer, more often at the custody layer.
Engine inverts the three things every other bot has gotten wrong.
This isn't a feature comparison. The differences below aren't about which dashboard looks nicer or whose model is bigger. They're about three structural choices that change what running a bot actually means for you — where your money sits, what you can see, and whose strategy is actually running.
If you've been considering automated trading and the offerings all start to blur together, the three differences below are the lens. Once you see them, you can't unsee them.
You keep your funds
Most trading bots ask for custody. The flow is the same across the category: you send funds to an account they control, the bot trades from that account, and you withdraw what's left when you're done. The terms of service tell you about the risks. The marketing tells you not to worry about them.
The disaster history says otherwise. Every year, more money is lost to bots-holding-funds failures than to bots-trading-poorly failures. Telegram-bot exits. Exchange-integrated services that abruptly froze withdrawals. "AI" platforms that vanished overnight after a Discord pump-and-dump. The common thread isn't bad models. It's bad custody.
Engine doesn't ask for custody. Your funds stay in an account you control. The agent has authority to trade — it can place orders, close positions, even lose money on bad trades. It does not have authority to withdraw. There is no Engine wallet pool. There is no "deposit address" on our side. We never touch your funds, because we never can.
The practical consequence: in the worst case for Engine — our backend gets breached, the team disappears, an angry intern pushes a malicious deploy — the blast radius is bounded by bad trades you can stop in one click. It is not your money is gone.
This is the difference that matters most because it addresses the actual #1 historical risk of running a bot. Every other claim in this post is downstream of it. If a bot product fails this question, the rest of the questions don't matter.
You see every decision
Most bots show you a P&L number. The good ones add a trade list. None tell you why.
This is the black-box problem, and it creates a specific kind of helplessness. When the bot wins, you can't tell whether it was smart or lucky. When it loses, you can't tell whether the strategy is working as designed and just having a rough month, or whether it's quietly broken and burning money on a stuck rule. You can't intervene because there's nothing to push on, no explanation to argue with, no specific decision to edit. You either trust the box or turn it off.
Engine streams every decision in plain English. Every market read. Every consider-and-skip. Every entry. Every exit. Every fill. With reasoning attached — not a model-weight hash, not "the agent thought there was an edge," but a specific reason traceable to a specific rule in your strategy.
You can watch it in real time. You can scroll backward. You can ask "why did you skip that setup at 2pm" and get an actual answer: the rule that gated it, the conditions that didn't match, the next setup the agent is watching for. You can disagree, edit the rule, and see what changes the next time the agent runs.
Transparency turns the bot from a slot machine into a colleague you can argue with. That's not a feature; it's the difference between a tool you actually understand and a tool you hope is doing its job.
The strategy is yours, not theirs
Every trading bot has to answer one question: whose rules is the bot running? Most have the same answer: ours.
The category splits into two shapes that both fail the same way. The "configurable" bots give you a dropdown of preset strategies — momentum, mean reversion, grid trading, DCA. You pick one and tune the parameters. The "AI" bots skip the dropdown and tell you the model figured it out. Either way, the rules belong to the bot owner. You're not running your strategy; you're renting access to theirs.
This isn't just about who owns the intellectual property. It's about who owns the edge. If the bot owner keeps the strategy, the upside concentrates in their hands — the better the strategy performs, the more they can charge, the more they can change the deal whenever they want. You're a subscriber, not an operator.
Engine inverts this. Your strategy is a plain-English file. You write it, or you fork one that's been published, and the agent reads it. The rules are yours: when to enter, what to size, where to stop, what to skip. You can keep the file private. You can publish it to the marketplace and earn from adopters. You can fork someone else's strategy, modify the parts you disagree with, and run it as your own.
The agent isn't your edge. The strategy is. The agent is the execution layer — the thing that reads the file around the clock and never gets tired, distracted, or scared. The edge stays with you, where it belongs.
You're not subscribing to a strategy. You're operating one.
Three inversions, one product
Your funds stay yours. Every decision is visible. The strategy is one you wrote — or one you can read before adopting.
If you've been running a bot and it stopped feeling like trading and started feeling like gambling, the three differences above are why. The shape of the category is what's broken, not the specific bot you chose.