Alex.arb

Founder SypherScore, @Vesper_mm
Building AI-powered funding rate arbitrage scanner

[Alex.arb @Cryptobalp]

@Xenon7889 @entropyIO This is just the beginning

[Alex.arb @Cryptobalp]

How I built ATMOS and took market making on @entropyIO to a whole new level.

Six months ago I thought market making was simple. Put orders on both sides of the price and earn the difference.

The market showed me how wrong I was.

Where it started

My first strategy on Entropy worked. Kind of. Volume was flowing, orders were getting filled. But every evening I looked at the numbers and saw the same thing. On a calm market I made pennies. On one sharp move I lost everything I had made that day.

So I started reading.

The first book was "Trading and Exchanges" by Larry Harris. One idea from it changed everything for me. In market making, what matters is not how often you trade. What matters is who trades with you. When your order gets filled right before the market moves, that is not luck. Someone on the other side knows more than you.

The second was the work of Easley, López de Prado and O'Hara on toxic order flow. They showed that this kind of flow leaves traces, and you can see them before it hits you.

The third was "Trades, Quotes and Prices" by Bouchaud and co-authors. It taught me how the order book really works and how every trade moves the price.

After those books I knew I didn't need a strategy that just places orders. I needed one that knows when not to.

Step 1. Data

For the first weeks I didn't write any strategy at all. I only collected data. Every trade, every order book change, every move on other venues, every one of my fills. Millions of events.

Then I went through my trades one by one. Why did this one go well and that one cost me? What was happening a second before it? Ten seconds before?

Step 2. The first patterns

The expensive trades were not random. They had a signature.

I saw that the market often moves in one place first and only then in another. If you watch where the price changes first, you can see the move before it reaches your order.

I saw that the most dangerous moment is right after a sharp move, while the market is still settling.

And I saw something I didn't expect at all. When the US session opens, the market becomes a different market. The spread behaves differently, the flow gets sharper, there are more professionals. What worked perfectly at night started costing money during US hours.

Step 3. I was getting in my own way

This was the biggest surprise. When many people run the same strategy, their orders naturally pile up at the same price. One big player could take them all at once, and those were the most expensive trades of all.

So I taught ATMOS to see all of its users at once and place them so they don't get in each other's way.

Step 4. Not fooling myself

The hardest part was not finding ideas. It was not fooling myself.

I had dozens of ideas that looked brilliant on paper. On the live market half of them didn't work. Some made things worse.

Once I spent a whole day celebrating a huge improvement. Then I found out it was just noise. The same "effect" showed up when I compared the strategy with itself.

After that I set a hard rule. No change reaches users until it proves itself on the live market, in a fair comparison, against a success criterion I write down in advance. First a test on historical data, then a live test, then a decision. No "let's see".

It slowed me down. It also saved me.

Step 5. The turning point

Over time one thing became clear. Any number you set by hand eventually becomes wrong. The market changes and the number doesn't.

Morning and evening, US and Asia, calm and storm. These are different markets, and each one has its own right answer.

So I stopped looking for the perfect settings and built a system that finds them on its own. ATMOS keeps testing, compares what works right now and shifts toward what works better. Every day. Without me touching it.

That is how a set of rules turned into a model that learns.

What came out of it

ATMOS sees the whole market at once. It picks its own place in the order book, decides when to enter and when to wait, and steps aside before a sharp move reaches your order. And it adapts to every market regime on its own.

How exactly it does that, I'm keeping to myself. That's months of work and the thing that sets it apart.

But I can say this. Every trade with ATMOS costs users less than anything I built before it. And it gets better every day.

ATMOS is live on Entropy. This is only chapter one.

[Vesper @vesper_mm]

The future starts right now.

Meet ATMOS - the first engine for @entropyIO and the most advanced machine learning model we've ever built.

It anticipates the market, adapts to any condition, and evolves with it in real time.

Months of development. Millions of market events. One goal - trading on a whole new level.

This is only the beginning.

[Alex.arb @Cryptobalp]

@0xmati_ @OndoPerps @vesper_mm Thanks bro

[Alex.arb @Cryptobalp]

80,643 referral points on @OndoPerps Is that a lot? 😄

And thanks to @vesper_mm for making it possible to trade on Ondo with low CPM - great to see our project helping users save on trading costs.

[Ondo Perps @OndoPerps]

The Ondo Points program now rewards Ondo Perps referrals.

Refer traders and earn referral points equal to 5% of the trading points they earn, with no cap on how many traders you can refer.

→ Retroactive to the start of the points program, so past referrals count.

→ Ongoing through the end of Season 1.

All referral points are in addition to the weekly 5M points drops for Perps traders. The retroactive points drop today.

Check your points → https://t.co/KqHZTwxJxp

[Alex.arb @Cryptobalp]

@maybach_eth @OndoPerps @vesper_mm I spent months optimizing Vesper to achieve a CPM below 100)

[Alex.arb @Cryptobalp]

@HqDefi44815 @OndoPerps @vesper_mm yes)

[Alex.arb @Cryptobalp]

@zilcxx @OndoPerps @vesper_mm Then I’ll share one with you.

[Alex.arb @Cryptobalp]

@zilcxx @OndoPerps @vesper_mm 1 Lambo?😅

[Alex.arb @Cryptobalp]

@Absoluto92 @OndoPerps @vesper_mm Thanks🥰

[Alex.arb @Cryptobalp]

@yakymoney @OndoPerps @vesper_mm Thank you, all of this is thanks to you.

[Alex.arb @Cryptobalp]

@0xmati_ Thanks, bro.

[Alex.arb @Cryptobalp]

Not saying anything. Just saying: make it before Sunday. Don't say I didn't warn you.

[Vesper @vesper_mm]

Vesper now has a home.

Our official Discord is live:

— bot updates
— new venues first
— direct line to the team

Join before Sunday and get the OG role.

Day one only, this window won't open again.

https://t.co/XxX6rgpkig

[Alex.arb @Cryptobalp]

@yakymoney Very big

[Alex.arb @Cryptobalp]

Half a billion. Built by a small team, one trade at a time.

Proud of this one.

[Vesper @vesper_mm]

Vesper just crossed $500,000,000 in volume on @OndoPerps

Half a billion, and we're only getting started.

One more thing: the Ondo team cut fees 30% for every Vesper user.

Our next target is under $100 per $1M of volume. Now it's within reach.

[Alex.arb @Cryptobalp]

@UneiduL Thanks bro

[Alex.arb @Cryptobalp]

2x for a week. Doing the math on my friends list right now.

[Vesper @vesper_mm]

What's better than a friend who trades?

A friend who earns you 2x.

All week, Vesper referral rewards are doubled. Invite your friends and earn twice as much from every trade they make.

Until October 21. After that, it's just friendship again.

https://t.co/JEGjKI9li7

[Alex.arb @Cryptobalp]

We generated nearly half a billion from Ondo alone.

[Vesper @vesper_mm]

$410,000,000.

That's how much volume we've done on @OndoPerps

Another $11M every day.

Ondo is now Vesper's most active market.

And our new mode, Aurum, runs it at 100 CPM:

just $100 in costs per $1M traded.

More volume. Less cost.

Turn on Aurum.

[Alex.arb @Cryptobalp]

Qfex and a new risk control core are now available on Vesper.

[Vesper @vesper_mm]

Wall Street just got a market maker.

@QFEX is now live on Vesper.

Perps on real-world assets, run by an engine that reads the market every tick and manages your position for you.

Live now: https://t.co/WcHHCxn20p

[Alex.arb @Cryptobalp]

Meet EXITE.

[Vesper @vesper_mm]

Meet EXITE.

A completely new strategy for @entropyIO by Vesper.
Rebuilt from scratch - from the first line to the last.

It sees the market. It decides on its own when to step in and when to wait.

Built with one goal: lower your CPM.

Months of development. Thousands of hours of testing on real data.

Our most advanced strategy to date. Live now.

https://t.co/RqZxRoAipg