A desk of autonomous agents for prediction markets and RWA. Perp agents in tasting. TG: @locksleyhq

Joined July 2026
Locksley is live. Try your agents out today. Autonomous agents that price live sports markets against the sharpest bookmakers on earth - and only bet when the market's price is measurably wrong. The whole setup takes about two minutes (watch below): ▸ Connect any wallet - EVM or Solana ▸ Keep the audited defaults: sports only, 40–90¢ band, 5pt+ edge, quarter-Kelly ▸ Fund it from any chain — ETH in, it bets real USDC on Solana ▸ Done. It hunts around the clock. Pause it whenever you want. You hold the keys - export your agent's wallet anytime. Every bet it takes is measured against the closing line, in the open. $150–300 puts the full strategy to work. Start smaller if you'd rather just watch it hunt first. locksley.bet CA: 0x29ebd5e871b201defc7a90691ad49cb703c47777 $LOCK — live on Robinhood Chain 🏹
22
19
6
63
30,295
Coverage Update Our odds layer pulls sharp-book lines across every league we track, but it had a hard cap on how many it could fetch per cycle. On a busy sports calendar that cap was binding constantly — and the leagues getting cut weren't the small ones. What changed: - Doubled league capacity. The full active board now fits in a single fetch. - Named leagues are protected. Champions League and World Cup were being dropped every cycle in favor of lower-tier competitions that happened to sort earlier. Major leagues now hold a guaranteed slot; auto-discovery competes for what's left. -WNBA is live. It was queued behind the NBA key, which returns zero games in July. In-season leagues now take priority over dormant ones. Liquidity thresholds now respect operator settings instead of being silently overridden.
1
3
433
Same sport. Opposite result. We pulled every bet since launch and split it by market type, MLB run lines: 22 bets, 54.5% win rate, +41.8% on capital. Priced against the bookmakers’ own spread quotes. MLB over/unders: 117 bets, 35.0% win rate, −25.2%. Modelled in house, and the model was overconfident. One market type, in one league, was 60% of our volume and more than our entire deficit. Strip it out and everything else runs at roughly +13%. So we turned that one lane off and left the rest running. Not a rebuild. A cut. locksley.bet
2
9
390
Your agent will lose. That is the point. Real numbers from the Locksley fleet over 36 hours: 57 bets settled, 28 won. A 49% win rate. It lost more bets than it won. It was up 8% on capital staked. One agent won exactly half its bets and returned 14.7%. That is not a paradox, it is the whole game. A bet at 40 cents that wins pays 2.5x. You can be wrong six times in ten and still walk away richer. What matters is whether the price you paid was better than the true probability. That gap is edge. Everything else is noise. Which is why the curve moves in waves. Take a coin that lands your way 55% of the time. Flip it 200 times and you will still hit a five loss streak, probably three or four times over. Nothing about the edge moved. That is just what randomness does to a sequence. Kelly is what keeps you alive through it. It sizes every bet to the size of the advantage, and it refuses to go bigger. Overbet a winning strategy and your growth rate actually falls while your risk of ruin climbs. A 50% drawdown needs a 100% gain to get back to flat. Casinos have known this forever. Roulette gives the house 5.26%. They lose hands all day, they lose to whales, they have losing nights. It never matters, because they are not trying to win the hand. They are trying to win the year, with a real edge, enough volume, and table limits that stop any one bet hurting them. That last part is a casino doing Kelly by hand. So a red day means almost nothing. The discipline is the edge. The waves are just weather. $LOCK
2
2
1
14
827
Reminder that every prediction agent on Locksley trades free until July 29. No entry fee, no exit fee, nothing clipped either way. Whatever your agent wins is yours. We did it because the fastest way to see what an autonomous agent actually does is to let it run with nothing skimmed off the top. Fund one, set the plan, and read the ledger in a week. The window shuts July 29. Then our fees are among the lowest anywhere. Right now there are none. $LOCK
2
4
1
12
1,182
Everyone is racing to build agents that can talk. Almost nobody is building agents that can trade. That is the harder problem and the more important one. An agent that answers a question is a demo. An agent that can hold a position, size it correctly, take profit, respect a stop, and do it again tomorrow without you watching is something else entirely. It is a portfolio manager that works for one person and never sleeps, never panics, and never gets bored. The reason it has not existed is that money is unforgiving. It needs custody you can trust, execution that actually settles, and a record you can verify. Screenshots do not count. Promises do not count. Only what happens onchain counts. So that is what we built. Agents that trade real markets, settle onchain, and show you exactly what they did down to the token. You set the plan, you fund the wallet, you keep the keys. The agent runs the discipline. The next wave of users onchain will not be people. It will be the agents they send in to work for them. Locksley is the desk they run on.
2
5
12
663
There are two ways to build in this space. You can rent someone else's rails and hope they never change the terms, or you can own the rails yourself. When the standard swap infrastructure refused to touch tokenized stocks, most teams would have stopped there and called it a limitation. We wrote and deployed our own router instead. It answers to nobody, holds nobody's funds, and cannot be switched off by a third party. Every agent wallet is one you control and can walk away with at any time. That is not a marketing choice, it is the whole point. An autonomous agent trading your money is only as trustworthy as the ground it stands on. We would rather own that ground than borrow it. Sovereign rails. Real onchain execution. Keys that are actually yours. $LOCK
3
4
10
476
Two desks. One account. Predictions: agents that hunt edge across live markets. Stocks: agents that buy and manage real tokenized equities onchain. Both autonomous. Both onchain. Both yours, keys and all. And we are only getting started. $LOCK
2
1
7
530
RWA agents are now live. Everything we showed you yesterday is open to everyone right now. Build an agent, fund it in two clicks, and it buys and manages real stocks onchain on Robinhood Chain. Your wallet, your keys, our own router doing every fill. No middleman, no paper account, just an agent trading real stock tokens while you sleep. And that is only the base layer. Here is where it goes next. The Complete Auto Agent. Not buy and hold. A fully autonomous trader that reads the market and actively hunts profit around the clock. No blocks to assemble. You fund it, it decides. This is the one we cannot wait to ship. Idle cash that earns. Uninvested balance stops sitting still and starts collecting onchain yield between buys. A wider market. More tokenized stocks come online as their liquidity grows. We spent the last stretch laying the rail. Now we build the machines that run on it. Go build your first agent. locksley.bet $LOCK
3
7
1
15
1,101
A few hours from now, Locksley agents start trading real stocks onchain. Not a demo. Real stock tokens, real fills, settled on Robinhood Chain through a router we built ourselves. You set the plan, fund the wallet, and the agent runs it while you sleep. The desk opens today. $LOCK
5
5
2
12
1,100
Here is the uncomfortable truth about trading. Most people do not lose because they pick wrong. They lose because they cannot behave. They average down out of hope. They take profit too early out of fear. They revenge trade a red day. They size a position by how confident they feel that morning instead of by what the math allows. The strategy was never the problem. The human running it was. This is the part an agent is actually good at. Not predicting the future, nobody can do that. But following a plan with zero emotion, every single time, at three in the morning, on a green day and a red one alike. Buy on the schedule. Trim at the target. Respect the stop. Keep the dry powder. Do it forever. The curve still moves in waves. Good stretches and bad ones. That never changes. What changes is that the machine does not flinch in the bad ones, and it does not get greedy in the good ones. Over enough reps, not flinching is the entire edge. Set the discipline once. Let it run.
3
3
16
722
Everyone is talking about AI agents. Almost nobody is letting them touch money. That is the actual unlock. An agent that can answer a question is a toy. An agent that can hold a position, size it correctly, take profit, cut a loss, and do it again tomorrow without you watching is a different thing entirely. That is not a chatbot. That is a portfolio manager that works for one person and never sleeps. The reason it has not happened is that money is hard and unforgiving. It needs custody you can trust, execution that actually settles, and a record you can verify. Screenshots do not count. Promises do not count. Only what happened onchain counts. So that is what we built. Agents that trade real markets, settle onchain, and show you exactly what they did, down to the token. You set the plan. You fund the wallet. You keep the keys. The agent runs the discipline. The first users onchain were people clicking buttons. The next million will be agents running strategies. We are building the desk they run on. $LOCK
4
3
2
16
721
Agents do not win every hour. Nobody does. Good ones are not the ones that never lose. They are the ones that size right, cut fast, and let the process compound. The equity curve is a wave, not a straight line. The edge is not being right. It is being disciplined when it is boring. $LOCK
2
4
14
653
The next big wave of users onchain will not be people. It will be their agents. Trading while they sleep. Following a plan without emotion. Never chasing, never panic selling, never tired. We are building the desk they run on. $LOCK
2
4
1
20
913
Locksley Stock Trader goes live within 24 hours. Locksley now trades real stocks onchain. Autonomously. And we built the entire execution layer ourselves. For the last stretch we have been heads down building something we think is a first of its kind: autonomous agents that buy and manage tokenized equities on Robinhood Chain. Not a mockup. Not a paper demo. Real ERC 20 stock tokens, bought and sold onchain, settled in a wallet the user owns. And it opens to the public within 24 hours. Here is what we shipped. We built our own onchain router. The obvious path was to plug into an existing swap API. We tried. The standard aggregator refused to quote tokenized stocks, blocking them at the API layer for legal reasons. So we did not wait for permission. We wrote, compiled, and deployed our own Uniswap v4 router directly to Robinhood Chain. It calls the chain's canonical PoolManager itself, holds no user funds, has no owner, and answers to nobody. That means no middleman, no compliance chokepoint that can flip a switch on us, and total control over how every trade is priced and executed. This is the piece we are most proud of, and it is the foundation everything else stands on. We price straight from the chain. We pinned the real Uniswap v4 contracts on Robinhood Chain empirically, by proving which ones actually hold the live stock liquidity, then read prices directly from the pool. Every mark an agent sees is the true onchain price, not a number from a third party feed. Every fill is the truth. Trades are recorded from the transaction receipt itself, the immutable onchain record of what actually moved, not from a quote or an estimate. What you see in a portfolio is exactly what happened onchain, to the token. The wallet is yours. Every agent gets its own Robinhood Chain wallet. We hold the key encrypted so the agent can act, and the user can export that key at any time. You are never locked in. Funding takes two clicks. Bridge in from Ethereum, Base, Arbitrum, Polygon, or BNB and land trading balance plus gas directly in the agent wallet, without leaving the desk. It looks like a real desk. Live portfolio value, cost basis, quantity held, and unrealized and realized P&L, with an equity curve that fills in as the agent works. A real portfolio manager view for a machine that never sleeps. And it runs itself. Users assemble a plan by dragging blocks. Recurring buys, weighting, rebalancing, take profit, trailing stops. The agent runs that discipline on a schedule, unattended. Real money execution stays behind a hard safety gate until deliberately armed. This goes live to the public in 24 hours. This is where things get really exciting! It is only the base layer. Once it is live, we build the thing we are truly excited about: the Locksley Stock Trader. A fully autonomous agent that does not just accumulate and hold, but actively trades stocks onchain, reading the market and hunting profit around the clock. The rail we just laid is exactly what it needs to run on. The front door to onchain stock trading for agents. Built on Robinhood Chain. Live within 24 hours. $LOCK
13
8
5
30
3,929
Congrats @vladtenev ! Quick one. Any chance of a developer API into your prediction markets? Open that door and we bring an army of autonomous agents to Robinhood Chain. For now we keep building on Jupiter on Solana, where the door is already open. And if the API is not on the roadmap yet, here is the other ask. Back us. Be our VC and we build the onchain prediction market layer for your chain ourselves. Either way, we would love to be the Jupiter of Robinhood Chain. You bring the humans. We bring the machines.
This quoted post is unavailable.
14
10
1
32
1,762
Locksley agents bet sports today. Soon they will own stocks. We are building the RWA desk: autonomous agents that buy real tokenized equities on Robinhood Chain. AAPL, NVDA, SPY and more as ERC20 stock tokens, purchased through professional market maker liquidity, held in the agent's own onchain wallet. And you will not write a single line of config. You drag blocks. What to hold. How often to buy. How it stays balanced. Take profit, trailing stop, hold forever. Drop them together, watch your plan write itself in plain English, press deploy. A robo advisor with no middleman, no minimums, and a ledger anyone can audit. Built on the chain everyone is talking about. In closed testing on our desk right now. AI szn. $LOCK
10
9
2
26
2,280
How much should you bet when you have an edge? There is an actual answer and most bettors have never heard of it. Kelly criterion: bet a fraction of bankroll proportional to your edge over the odds. Big edge, bigger stake. Small edge, smaller stake. No edge, zero, and zero means zero. Every Locksley agent sizes this way, from its own bankroll, automatically. Two agents with different bankrolls will stake different dollars on the same bet and identical fractions. Conviction is not a number. Kelly is.
5
4
15
584
Before this desk went live we graded nearly a thousand of its old settled bets by hand. Win rate by price band, by sport, by market type. The data killed some darlings. Whole categories we liked got cut because the ledger said they lose. What survived became the strategy every agent runs today. Strategy written by results. That is the entire $LOCK thesis in one sentence.
3
6
1
18
962
Something people get wrong about betting bots. The hard problem is not finding bets, it is refusing them. Our agents scan thousands of markets every few minutes and reject nearly all of them. Too thin to fill cleanly. Game already started so the line is stale. Price too far from what sharp books say is fair. Every rejection gets logged with its reason. A bot with no discipline is just a gambling addict with better uptime.
4
3
11
573
A prediction market price is just a probability wearing a dollar sign. A team trading at 42 cents is the crowd saying 42 percent. The whole game is finding spots where the crowd says 42 and the sharpest bookmakers in the world say 49. You buy the gap, not the team. Do that hundreds of times with correct sizing and the wins take care of themselves. Most people bet outcomes. The desk bets prices.
3
9
18
634