Starts in simulationMeasured against chanceHelp the network, if you want
The bot that proves it before it trades.
Bajamar runs on your computer, with your wallet. It measures every strategy against random entries and against doing nothing, and only what survives the rule moves on to real money. If you want, your bot helps the network with one button, and you switch it off whenever you like. What the network learns is free for everyone.
No system wins every day. Nobody here promises returns; we show what we measure.
Illustrative example, not a result
B · floor 95 %, m = 2
Few trades
+4.2 %+0.0 %+4.2 %
net on capital, 90 days
Trades
27027
vs. random
p91p0p91
Holding SOL
-3.1%-0.0%-3.1%
27 of 30 trades needed for a verdict
16016
strategy rules put through formal tests
none has cleared its bar
000
rules changed after seeing the data
every rule is written before the data, and the data judges it
> 5,000> 0> 5,000
automated tests
every change comes with its own, and all of them pass before anything goes out
Closed tests, on prices from 2024 to 2026, costs included. The past is not the future.
How it works
Three engines, each with what is known about it
There is no magic strategy. There are three engines, each measured separately, with what is known about it written right next to it.
Engine A
Capital works while it waits
Idle USDC is lent on Jupiter Lend and earns interest.
Watched all the time: that USDC is still worth a dollar, that there is money to withdraw and that nobody changes the rules of the program. If anything goes wrong, it takes the money out by itself.
Real, off by default
Engine B
How much SOL to hold, by the trend
Decides once a day by looking at the trend over nine different spans, never going below a capital floor: in a trend, in; in a crash, out.
On past data it earned less than plain lending in 6 out of 10 one-year periods. It is being measured in simulation.
In simulation, being measured
Engine C
Capitulation hunter
Buys a sharp drop on huge volume and sells within a few hours, or sooner if it keeps falling.
In the data it wins before costs; after costs, it is promising. That is why it stays in simulation.
In simulation, being measured
Everything starts in simulation; nothing touches real money without your permission.
The network
Your bot can help the network
If you want, your bot helps the network: it watches the coins it is given, at their set times, and reports what it sees. What the network learns is free and the same for everyone.
Danger warnings, on trial
The danger warnings between bots are already in the app, but they have not yet warned of any of the sudden drops the network has seen. We are looking into why and, meanwhile, we do not count on them. A warning never makes your bot buy.
What to avoid, on trial
The network prepares, the same for everyone, a list of what to avoid among new coins. The first one came out empty: we are reviewing the count, and it is switched on once it has been checked. It only warns: it never buys or sells for anyone.
You decide
It comes switched off and only goes on with your permission. Your bot uses a key of its own, never the wallet you trade with, and it does not send your keys or anything you do with real money. Your Solana address is optional. Switch it off whenever you like.
Next: the more bots are helping, the more of them watch each coin, each at a different moment, so that together they catch the sudden drops that today slip by between two looks. It comes in when it passes its tests, and not before.
Points
Points have no value and no use, and cannot be bought, sold or exchanged. They only show how much you have helped the network.
Learning
The rule is written before the data is seen
Every strategy is compared with two unforgiving benchmarks: holding SOL, and itself with random entries. If it does not beat both, the signal adds nothing.
1
≥ 30≥ 0≥ 30
trades closed by the strategy itself, over 10 days or more
2
95 %0 %95 %
better than 95 out of 100 versions that buy at random
3
> SOL
beats just holding SOL over the same dates
4
> 0> 0> 0
on average, after costs and with its day-by-day margin of error
What is measured, and how
1
Preregistered variants
Each registered variant is written before seeing results and never tweaked along the way. Since registration, everything tested is counted, including what fails; for what was looked at before registering, we say what is known.
2
Two benchmarks
Holding SOL over the same dates, and the strategy itself with random entries (300 versions).
3
One rule
An edge only with 30 trades over 10 days, a mean above 0 with its own day-by-day error, everything tried in its family counted, and beating random entries and holding SOL. To move to real money, it also has to beat plain lending.
4
Decision journal
Why the bot did or did not do each thing, 90 days, exportable. Without causes there are only results.
Five verdicts, the same rule for all
Shown in the app's panel.
No tradesFew tradesNo edgePromisingProven edge
What we look at for each strategy
how often it gets it right, how much it wins against what it loses, what each trade leaves on average, how sure that result is, the worst drop, time in the market and streaks
What is never done
turning a few dozen trades into a yearly figure, changing numbers along the way, counting what you close by hand
Security
What protects your money, point by point
A bot that trades with your wallet has to be able to explain itself. This is what exists today, without decoration.
Your key never leaves your computer
The wallet is stored encrypted with your password and the bot runs on your computer. Nobody at Bajamar sees it. And the app only installs updates that come from us: it checks before installing.
Simulation first, always
Everything starts in simulation. Real money comes switched off and is only switched on deliberately, with your password. Before every real trade, the bot rehearses it and compares it with what it expected: if it does not match, it does not go ahead.
Loss brake
There is a daily loss limit. If it is hit, the bot stops buying and keeps its exits ready to sell. It only buys again when you say so.
The network does not touch your money
The network never buys, sells or changes amounts for you. A warning from the network only makes your bot look again for itself, and it never makes it buy.
Nobody will ask for your keys
Nobody from the project will ever ask you for your secret phrase or your keys, or ask you to send money to any address. There is no Bajamar token on the real network: if you see one for sale, it is a copy made to deceive you.
Security flaws, in private
If you find a flaw, write to us privately at [email protected]. Do not make it public. We reply within 72 hours.
News
Latest from Bajamar
What has changed in the app and in the network, in plain words.
Mac (.dmg) or Windows (.exe), from the download page.
2
Install it
Mac: drag it to Applications. Windows: double-click the installer. After that it updates itself.
3
Create your wallet and start
The panel opens: pick a password, save your words and turn on the engines in simulation.
Mac: Intel and Apple Silicon Windows 10 and 11
Questions worth asking
Does it make money?+
With the original strategies, no: the audit measured it on real data and says so. Today only lending USDC gives an almost certain return, with the risk of the protocol. The trend with a floor and the drop hunter are unproven: they are measured in simulation, against plain lending. Everything starts in simulation and is judged by the same rule.
What about new coins?+
They are for learning what to avoid, not for making money. The bot only touches them in simulation, as a measurement, and with today's data no way of picking them returns more than was put in. In new coins, out of every 100 $ put into 152 simulated trades closed from 25 to 28/09/2026 (138 with the old rule and 14 with today's), 72 $ came back, with the costs the simulation measures; with the account rent, which the simulation did not charge back then, 69 $. In 32 of them more than three quarters of what was put in was lost. Doing nothing keeps those 100 $ at 100 $; plain lending pays 3.6 to 4.6 % gross a year, with the protocol's risk. Measuring goes on, and if nothing beats not trading, that is where they stay: what to avoid.
Do I need to buy MAREA to use the bot?+
No. MAREA is not sold. The app and simulation are free.
What do I get if I help the network?+
Points. Points have no value and no use, and cannot be bought, sold or exchanged. They only show how much you have helped the network. The danger warnings between bots are already in the app, but they are still on trial and we do not count on them. What the network learns, such as the list of what to avoid, which is also on trial, is free for everyone, whether you help or not.
Where are my keys?+
On your computer, encrypted. If you help the network, your bot sends what it sees and decides in simulation, with a key of its own that is not your wallet. Never your keys, the wallet you trade with or what you do with real money. Your Solana address, only if you add it yourself: it is optional.
Where can I use it?+
Everywhere in the world except mainland China and countries under sanctions. Not available in mainland China. Nor in Russia, Belarus, North Korea, Iran or the areas of Ukraine occupied by Russia. You must be 18, or the age of majority in your country if it is higher.
When does the token come out?+
There is no date and no promise. Today MAREA only exists on the test network, where it is worth nothing, and the Bajamar network gives it to nobody. On the real network there is no MAREA token, nor any other token related to Bajamar: if you see one for sale, on any website, exchange or social network, it is a copy made to deceive you. We will never ask you for your secret phrase or ask you to send us cryptocurrency.