Crypto Arbitrage in 2026: What a Retail Quant Can Still Do

There is a fairly common misconception that proper HFT, arbitrage, and strategies that are sensitive to latency and execution quality are now exclusively the domain of large funds.

In traditional financial markets, that is largely true.

But crypto works somewhat differently.

There are still a huge number of independent exchanges, markets, instruments, and execution environments. Every venue has its own fees, APIs, liquidation rules, order book structure, pricing formulas, and participants.

That is exactly why some arbitrage strategies can still be run by a single person.

You don’t necessarily need a hedge fund, a team of twenty quants, and a server room in New York. In some cases, a decent computer, several APIs, a VPS, and well-written code are enough.

Of course, the market has become much more competitive. The most obvious opportunities have been picked over long ago. Competing in the largest and most liquid markets against professional firms is usually pointless.

But that does not mean arbitrage is dead.

It has simply moved toward places where it is not economically interesting for the biggest players to operate.

Which strategies are still accessible

This list is far from exhaustive. But even today, there are several areas that can potentially be approached by a small quant project:

• Funding Rate Arbitrage
• Liquidation Arbitrage
• Mark Price Arbitrage
• Cross-Exchange / Triangular Arbitrage
• Queue Position / Speed Arbitrage
• Spot-Futures Arbitrage
• Options Pricing Arbitrage

Why does this still work at all?

The main reason is very simple.

The crypto market is still fragmented.

Instead of having a handful of major trading venues, like traditional financial markets, crypto has a huge number of exchanges and trading systems. They operate independently and differ in almost everything:

• fees;
• APIs;
• order processing speed;
• liquidation rules;
• order book structure;
• funding mechanics;
• price calculation methods;
• liquidity;
• available instruments.

This creates a rather strange situation.

On one hand, crypto is becoming more efficient.

On the other hand, there are so many venues and instruments that making the entire market perfectly efficient is practically impossible.

A large player can spend huge amounts of money and manpower optimizing a single market.

But if that market can only generate a few thousand dollars a day, hiring a $200k-per-year quant to work on it simply does not make economic sense.

For a small algorithmic project, the economics are completely different.

If you can build the system yourself, the main cost is your own time.

And this is exactly where the window for the retail quant appears.

But there is an important catch: don’t go where everyone else is

Market selection is often more important than the strategy itself.

For example, trying to compete on BTC/USDT on the largest exchange is probably not a great idea.

The largest crypto trading firms and traditional financial players are active in these markets, bringing all of their infrastructure capabilities with them.

They have better networks, better execution, more capital, and more people.

Trying to beat them simply because you have a good Python script probably won’t work.

It is much more interesting to look for less saturated markets.

For example:

• a small-cap token on a major exchange;
• BTC/USDT on a less popular venue;
• a rare derivative;
• a market with unusual mechanics;
• an instrument that is simply too small for major firms to care about.

And here we get an interesting paradox.

The smaller the market, the less liquidity it has.

But at the same time, there may also be less competition.

For a hobby quant, that can be much more important.

1. Funding Rate Arbitrage

This is probably the most well-known crypto arbitrage strategy.

The idea is extremely simple.

The funding rate of one perpetual contract is different from the funding rate of another.

You structure positions so that you receive a high funding rate on one venue and pay a lower funding rate on another.

In the ideal case, the resulting position is almost delta-neutral.

But there is an obvious problem: funding rates constantly change.

Something that looked profitable five minutes ago may no longer be profitable after the next funding calculation.

You also need to account for:

• fees;
• spread;
• slippage;
• entry costs;
• exit costs;
• funding changes;
• liquidation risk;
• exchange risk.

So simply writing a bot that looks for the highest funding rate is not enough.

You need to calculate the actual expected return of the entire position structure.

There is also a more exotic version: funding arbitrage between perpetual options and perpetual futures.

Funding on some perpetual options can be significantly higher than on conventional perpetual futures.

And these unusual markets are often where a small quant project can become more interesting.

2. Spot-Futures Arbitrage

Another classic approach is cash and carry.

You buy the underlying asset and simultaneously sell the corresponding futures contract.

If the futures contract trades at a sufficient premium, or the perpetual has a sufficiently high funding rate, you can gradually capture the difference.

The key is to keep the overall position neutral to the direction of the market.

If BTC goes up 20%, it shouldn’t matter much to you. If it drops 20%, it shouldn’t matter either.

You are not trying to make money from the movement of BTC itself. You are making money from the difference between two instruments.

With dated futures, you can hold the position until expiration.

With perpetual contracts, funding becomes the source of the return.

And this is where the strategy can become particularly interesting during strong bull markets.

During periods of extreme market enthusiasm, funding on major coins can become very high.

For example, an average funding rate of around 0.3% per day, compounded daily, corresponds to roughly 298% annualized.

Of course, this does not mean such a return can be maintained for an entire year. Funding changes, markets change, and extreme funding usually appears precisely during periods of overheating.

There are also obvious risks:

• insufficient collateral;
• liquidation;
• funding changes;
• fees;
• counterparty risk;
• exchange failure or shutdown.

3. Mark Price Arbitrage

This is where things start getting more interesting.

Many exchanges use a Mark Price — an artificial price that is used, among other things, to determine liquidations.

The exchange calculates it according to a specific formula and sends it to clients through its API.

On some implementations, updates may occur every 100 milliseconds, for example.

But if the calculation formula is public, a simple question arises:

why wait for the exchange to calculate the next Mark Price?

You can calculate it yourself.

And on modern hardware, this kind of calculation typically takes less than a millisecond.

That creates a small informational advantage.

You can obtain your own estimate of the next Mark Price before some market participants receive the next update from the exchange itself.

By itself, this is not a complete trading strategy.

But it can be a source of alpha.

Especially when combined with a market-making or scalping system that can directly convert a small informational advantage into trades.

4. Cross-Exchange and Triangular Arbitrage

The most obvious idea is that the same asset can have different prices on different exchanges.

Buy cheaper.

Sell more expensive.

In theory, everything looks perfect.

In practice, this particular type of arbitrage is already quite saturated.

That is why looking for huge obvious price differences between the largest exchanges usually makes little sense.

But smaller markets and triangular arbitrage opportunities still exist.

At this point, infrastructure becomes just as important as the mathematics.

You need to minimize:

• network latency;
• data processing time;
• signal calculation time;
• order submission time;
• execution latency;
• slippage.

The classic idea of co-location also does not always apply in exactly the same way as it does in traditional markets.

Most crypto exchanges operate through cloud infrastructure.

So what matters much more is being in the right data center relative to the specific exchange and having an efficient network and software stack.

There is also another unpleasant issue: fees.

Even if a strategy is mathematically profitable, fees can consume a significant portion of the result.

The larger your trading volume, the better your fee tier usually becomes.

But for a small trader, a substantial part of the theoretical profit can simply end up back at the exchange as trading fees.

5. Liquidation Arbitrage

Liquidations are an interesting source of forced flow.

When a trader’s position becomes insolvent, the exchange has to close it.

And it is not doing so because it believes the price is attractive.

It simply needs to get rid of the position.

As a result, a liquidation engine can generate aggressive market or IOC orders.

And an aggressive forced order can move the order book significantly.

If an algorithm can detect the beginning of such a liquidation quickly enough, it may be possible to trade against the resulting move.

In a sense, this is a fairly pure market-microstructure strategy.

You are not trying to predict where Bitcoin will be in an hour.

You are trying to understand what the liquidation mechanism of a specific exchange is doing right now.

And use that as a source of short-term edge.

6. Speed / Queue Position Arbitrage

This is already getting close to HFT, but there is an important detail.

You do not necessarily need to be the fastest participant in the entire market.

You only need to be fast enough in a specific small market.

There is a queue of orders inside an order book.

If you can consistently obtain a favorable position in that queue, the quality of your fills changes.

Instead of receiving fills randomly, you can potentially get more trades with positive expected value.

Different techniques can be used to achieve this.

For example, maintaining orders across multiple price levels or exploiting an advantage in the speed of market-data processing and order submission.

This type of mechanism can be combined with practically everything else:

• cross-exchange arbitrage;
• funding arbitrage;
• liquidation arbitrage;
• market making;
• scalping.

And here we get back to the main principle of retail HFT:

you don’t have to be the fastest in the world. You only have to be fast enough where everyone else isn’t.

If a large player considers a market interesting only once it reaches $10M in daily volume, a $2M-per-day market may have practically no major competitors.

Suppose an algorithm can capture only 2 bps on traded volume.

At $2M of daily volume, that is roughly $400 per day.

At $5M, it becomes roughly $1,000.

Of course, this is a very rough calculation before accounting for all real-world costs and is by no means a guarantee of profitability.

But this is exactly the scale at which the economics of a hobby quant can start becoming interesting.

7. Options Pricing Arbitrage

Crypto options are still significantly less developed than options markets in traditional finance.

And that automatically means there can be more pricing inefficiencies.

While options represent a huge part of derivatives trading in traditional finance, their share in crypto is still significantly smaller.

That means the market is not yet as efficiently priced.

Exotic instruments are particularly interesting.

For example:

• perpetual options;
• basket options;
• other non-standard derivatives.

Here, the problem is not necessarily speed. It is the model.

If you have a model that estimates the fair value of an option better than the market does, you can potentially buy the undervalued instrument and sell the overvalued one.

This is much closer to traditional quantitative trading.

The advantage may not be measured in milliseconds at all.

It may be entirely in the mathematics.

But there is an unpleasant side as well.

Options can have extremely high fees.

Some venues can charge double-digit percentages of notional value.

Therefore, the fee structure of a particular exchange becomes part of the mathematical model itself.

So why choose only one strategy?

This is probably the most interesting part.

There is no need to build seven independent bots.

You can build one trading infrastructure and connect several sources of alpha to it.

For example:

• collect market data from multiple exchanges;
• calculate funding rates;
• calculate your own Mark Price;
• monitor liquidations;
• compare order books;
• calculate cross-exchange opportunities;
• manage order queues;
• execute trades through a unified execution engine.

At that point, you are no longer building an “arbitrage bot”.

You are building a small trading system.

The same infrastructure layer can support multiple strategies.

The strategy itself becomes essentially a collection of signals and risk-management rules.

What actually makes sense in 2026?

If we remove all the marketing noise around the word HFT, the situation looks roughly like this:

And this is the important part

Crypto arbitrage is not dead.

What is dead is the idea that you can simply open two exchanges, spot a price difference, and easily take the money.

Those opportunities have indeed become rare.

But the market has simultaneously become much more interesting from an engineering perspective.

Today, the edge may not be in an obvious price difference at all.

It may be hidden in the way a particular exchange calculates its price, how its liquidation engine works, how funding is updated, how its order book behaves, how queue priority is determined, or how quickly your code can process an event.

And this is exactly why small quant projects can still have a place.

Not because large players are stupid.

But because the market is so large and fragmented that it is simply not economically rational for them to optimize everything.

Your job as a small quant is not to beat Citadel, Jane Street, or the next crypto HFT fund on their own territory.

Your job is much simpler:

find a market that is too small, too strange, or too inconvenient for them — and build a system that understands it better than everyone else.

And that is where things start getting really interesting.