❓What is HETHA
HETHA is built as a full-cycle arbitrage system.
It collects market data, builds depth-2/3 arbitrage chains, filters them, and sends the valid ones to the HETHA trading bot, which executes trades according to your settings or runs in simulation.
The full workflow looks like this:
data collection → chain building → filtering → HETHA bot → result.
This is the foundation everything else in HETHA is built on.
@hetha_io
HETHA is built as a full-cycle arbitrage system.
It collects market data, builds depth-2/3 arbitrage chains, filters them, and sends the valid ones to the HETHA trading bot, which executes trades according to your settings or runs in simulation.
The full workflow looks like this:
data collection → chain building → filtering → HETHA bot → result.
This is the foundation everything else in HETHA is built on.
@hetha_io
📎 What are arbitrage chains?
In HETHA.IO, every arbitrage opportunity is packaged as a chain — a short execution route between exchanges that must run almost simultaneously.
There are two types:
🔹 Depth-2: simple buy→sell across two exchanges (e.g., bnc:eth/usd-bbt:eth/usd)
🔹 Depth-3: a short triangular execution loop (e.g., bnc:btc/usd-kcn:eth/usd-bnc:eth/btc)
The system continuously evaluates large numbers of chains and keeps only those that remain realistically executable under live market conditions.
Executed arbitrage chains (screenshot).
@hetha_io
In HETHA.IO, every arbitrage opportunity is packaged as a chain — a short execution route between exchanges that must run almost simultaneously.
There are two types:
🔹 Depth-2: simple buy→sell across two exchanges (e.g., bnc:eth/usd-bbt:eth/usd)
🔹 Depth-3: a short triangular execution loop (e.g., bnc:btc/usd-kcn:eth/usd-bnc:eth/btc)
The system continuously evaluates large numbers of chains and keeps only those that remain realistically executable under live market conditions.
Executed arbitrage chains (screenshot).
@hetha_io
🎯 Example: 2-Leg Chain Walkthrough
A 2-leg chain is the simplest form of arbitrage: it starts and ends in BTC, converting BTC to ADA on one exchange and converting ADA back to BTC on another.
Example: bfx:ada/btc-bnc:ada/btc
Before execution, the system checks:
🔹 that profit remains positive after fees,
🔹 that prices still match live orderbooks,
🔹 and that the chain is fresh enough to execute.
If a chain becomes too old, market conditions may already have changed.
When all conditions match the current orderbooks, the chain is sent for execution.
@hetha_io
A 2-leg chain is the simplest form of arbitrage: it starts and ends in BTC, converting BTC to ADA on one exchange and converting ADA back to BTC on another.
Example: bfx:ada/btc-bnc:ada/btc
Before execution, the system checks:
🔹 that profit remains positive after fees,
🔹 that prices still match live orderbooks,
🔹 and that the chain is fresh enough to execute.
If a chain becomes too old, market conditions may already have changed.
When all conditions match the current orderbooks, the chain is sent for execution.
@hetha_io
🖇 Example: 3-Leg Arbitrage Chain
A 3-leg chain is a sequence of three actions across different markets that returns to the same base currency.
Example: bnc:sol/btc-bbt:sol/usd-bnc:btc/usd
The chain starts in BTC. On the first exchange, BTC is converted into SOL. On the second market, SOL is converted into USD. On the final step, USD is converted back into BTC. The opportunity exists only while all three conversions remain aligned after fees.
The system verifies that the result remains positive after fees, that weighted prices still match the live orderbooks, and that liquidity is sufficient across all three steps.
Because a depth-3 chain depends on three markets, even small changes can affect its result.
@hetha_io
A 3-leg chain is a sequence of three actions across different markets that returns to the same base currency.
Example: bnc:sol/btc-bbt:sol/usd-bnc:btc/usd
The chain starts in BTC. On the first exchange, BTC is converted into SOL. On the second market, SOL is converted into USD. On the final step, USD is converted back into BTC. The opportunity exists only while all three conversions remain aligned after fees.
The system verifies that the result remains positive after fees, that weighted prices still match the live orderbooks, and that liquidity is sufficient across all three steps.
Because a depth-3 chain depends on three markets, even small changes can affect its result.
@hetha_io
📈 Order book Basics
Arbitrage always starts with the order book. It has two core sides:
🟢 Bid — prices buyers are willing to pay
🔴 Ask — prices sellers are offering
The difference between them — the spread — determines whether an arbitrage opportunity exists at all.
On the chart, bids are on the left (green), asks are on the right (red). Even a small shift between them can determine whether a chain remains profitable.
That’s why order books matter more than “prices” in isolation.
@hetha_io
Arbitrage always starts with the order book. It has two core sides:
🟢 Bid — prices buyers are willing to pay
🔴 Ask — prices sellers are offering
The difference between them — the spread — determines whether an arbitrage opportunity exists at all.
On the chart, bids are on the left (green), asks are on the right (red). Even a small shift between them can determine whether a chain remains profitable.
That’s why order books matter more than “prices” in isolation.
@hetha_io
📉 This is what arbitrage looks like in practice
Not “buy here, sell there”. A log of executed chains.
Each row in the table is an arbitrage chain that passed liquidity checks, fees, and price freshness — and was actually executed, not just found by calculation. Only a small fraction of all detected opportunities ever make it into this table.
Most “profitable” chains die earlier — at the market validation and execution stage.
This is where theory ends and real arbitrage begins.
@hetha_io
Not “buy here, sell there”. A log of executed chains.
Each row in the table is an arbitrage chain that passed liquidity checks, fees, and price freshness — and was actually executed, not just found by calculation. Only a small fraction of all detected opportunities ever make it into this table.
Most “profitable” chains die earlier — at the market validation and execution stage.
This is where theory ends and real arbitrage begins.
@hetha_io
📹 Video: Inside HETHA — Core Pipeline
This video gives a clear overview of the core pipeline behind HETHA. It walks through the same structure we introduced at the beginning:
data → chains → filtering → execution.
Throughout the last two weeks, we looked at individual parts of this pipeline — arbitrage chains, depth, validation. This video puts everything back together into a single system view.
This is the architectural baseline everything else builds on.
Watch: https://www.youtube.com/watch?v=7kCh59rphWE
@hetha_io
This video gives a clear overview of the core pipeline behind HETHA. It walks through the same structure we introduced at the beginning:
data → chains → filtering → execution.
Throughout the last two weeks, we looked at individual parts of this pipeline — arbitrage chains, depth, validation. This video puts everything back together into a single system view.
This is the architectural baseline everything else builds on.
Watch: https://www.youtube.com/watch?v=7kCh59rphWE
@hetha_io
🌊 Tick Flow: how HETHA processes 300k+ updates per minute
Every tick is a change in price or volume on an exchange. The system processes around 100,000–300,000 ticks per minute.
How it works:
🔹 Workers receive ticks from exchanges, maintain orderbooks, and keep them up to date.
🔹 Ticks that do not modify key orderbook parameters (aW, bW, aB, bB) or change them by less than 0.01% are ignored to avoid unnecessary recalculations.
🔹 If a change is meaningful, the updated orderbook is sent to the GPU server, where chains are recalculated.
🔹 Out of the thousands of chains recalculated every hour, only those that remain realistic and aligned with the live market survive.
This is how HETHA validates arbitrage chains in real time.
@hetha_io
Every tick is a change in price or volume on an exchange. The system processes around 100,000–300,000 ticks per minute.
How it works:
🔹 Workers receive ticks from exchanges, maintain orderbooks, and keep them up to date.
🔹 Ticks that do not modify key orderbook parameters (aW, bW, aB, bB) or change them by less than 0.01% are ignored to avoid unnecessary recalculations.
🔹 If a change is meaningful, the updated orderbook is sent to the GPU server, where chains are recalculated.
🔹 Out of the thousands of chains recalculated every hour, only those that remain realistic and aligned with the live market survive.
This is how HETHA validates arbitrage chains in real time.
@hetha_io
📊 Full Stream vs Filtered Stream — what’s the practical difference
In HETHA, there are two ways to observe arbitrage chains: Full Stream and Filtered Stream.
🔹 Full Stream shows the entire flow of chains generated from real-time market data, without applying any post-generation filtering. It is useful for analysis, diagnostics, and experimentation.
🔹 Filtered Stream shows the same generated chains, but only those that pass a configurable set of post-generation filters, such as profitability thresholds, generation time limits, and acceptable order book deviation. It is designed for working with chains that pass the current filtering criteria and are suitable for further use under current market conditions.
Both streams rely on the same underlying data and calculation logic. The difference is not in the algorithm itself, but in whether filtering is applied and which subset of chains is displayed.
The screenshot above shows how the Full Stream looks in practice.
@hetha_io
In HETHA, there are two ways to observe arbitrage chains: Full Stream and Filtered Stream.
🔹 Full Stream shows the entire flow of chains generated from real-time market data, without applying any post-generation filtering. It is useful for analysis, diagnostics, and experimentation.
🔹 Filtered Stream shows the same generated chains, but only those that pass a configurable set of post-generation filters, such as profitability thresholds, generation time limits, and acceptable order book deviation. It is designed for working with chains that pass the current filtering criteria and are suitable for further use under current market conditions.
Both streams rely on the same underlying data and calculation logic. The difference is not in the algorithm itself, but in whether filtering is applied and which subset of chains is displayed.
The screenshot above shows how the Full Stream looks in practice.
@hetha_io
🌐 How the GPU decides which chains survive — and which ones “die”
Once the order books are updated, the most interesting stage begins: chain recalculation.
HETHA does not store ready-made routes. Arbitrage chains are recalculated from scratch every time the market structure changes.
At this stage, the GPU checks:
🔹 whether each step of the chain still matches the current order books,
🔹 whether profitability remains positive after fees are applied,
🔹 whether there is sufficient liquidity at every step,
🔹 whether any abnormal values appear after recalculation.
If a chain stops matching the market even at a single step, it is discarded instantly.
This is what real arbitrage is: a chain “lives” only as long as the market continues to confirm it.
@hetha_io
Once the order books are updated, the most interesting stage begins: chain recalculation.
HETHA does not store ready-made routes. Arbitrage chains are recalculated from scratch every time the market structure changes.
At this stage, the GPU checks:
🔹 whether each step of the chain still matches the current order books,
🔹 whether profitability remains positive after fees are applied,
🔹 whether there is sufficient liquidity at every step,
🔹 whether any abnormal values appear after recalculation.
If a chain stops matching the market even at a single step, it is discarded instantly.
This is what real arbitrage is: a chain “lives” only as long as the market continues to confirm it.
@hetha_io
📊 Order books: how market microstructure affects arbitrage
When we talk about arbitrage, everything ultimately comes down to the order book. Not the “price on the screen” — but how volume is distributed across bids and asks.
Even if a price difference looks attractive, insufficient depth or an unstable spread can prevent a chain from remaining valid during recalculation. That’s why chains are evaluated against the current state of the order book, not against a single quoted price.
Arbitrage is not a number. It is the interaction between volume, spread, and timing.
@hetha_io
When we talk about arbitrage, everything ultimately comes down to the order book. Not the “price on the screen” — but how volume is distributed across bids and asks.
Even if a price difference looks attractive, insufficient depth or an unstable spread can prevent a chain from remaining valid during recalculation. That’s why chains are evaluated against the current state of the order book, not against a single quoted price.
Arbitrage is not a number. It is the interaction between volume, spread, and timing.
@hetha_io
🖇 How GPU participates in arbitrage in HETHA
In HETHA, the GPU is not a “signal accelerator”. It is a computation layer that allows the system to continuously recalculate chains as the market changes.
Collector servers maintain orderbooks and pass updates further down the pipeline. When changes are significant, chains are recalculated — and this is where the GPU is used.
Most chains “die” during recalculation because the market no longer confirms them. The GPU is not needed to find more chains, but to ensure that only those aligned with the current market remain.
@hetha_io
In HETHA, the GPU is not a “signal accelerator”. It is a computation layer that allows the system to continuously recalculate chains as the market changes.
Collector servers maintain orderbooks and pass updates further down the pipeline. When changes are significant, chains are recalculated — and this is where the GPU is used.
Most chains “die” during recalculation because the market no longer confirms them. The GPU is not needed to find more chains, but to ensure that only those aligned with the current market remain.
@hetha_io
📉 What Is Chain Generation Time & Why It Matters
Chain generation time is the interval between the last relevant market tick and the moment the calculated chain becomes available in the system.
In a live market, order books change continuously. If generation time increases, the probability that the underlying market state has already shifted also increases.
For this reason, HETHA applies a generation time threshold. Chains exceeding the threshold are not considered unreliable — they are simply no longer guaranteed to reflect the current order book state.
Generation time is not about speed for its own sake. It is a measure of temporal relevance.
@hetha_io
Chain generation time is the interval between the last relevant market tick and the moment the calculated chain becomes available in the system.
In a live market, order books change continuously. If generation time increases, the probability that the underlying market state has already shifted also increases.
For this reason, HETHA applies a generation time threshold. Chains exceeding the threshold are not considered unreliable — they are simply no longer guaranteed to reflect the current order book state.
Generation time is not about speed for its own sake. It is a measure of temporal relevance.
@hetha_io
An arbitrage portfolio ≠ an investment portfolio
What matters is not how many assets you have, but which ones actually participate in chains.
What really matters:
🔹 liquidity
🔹 base currency alignment
🔹 share of used assets
🔹 how many chains survive until execution
In arbitrage, it’s not coin selection that works — it’s portfolio structure.
@hetha_io
What matters is not how many assets you have, but which ones actually participate in chains.
What really matters:
🔹 liquidity
🔹 base currency alignment
🔹 share of used assets
🔹 how many chains survive until execution
In arbitrage, it’s not coin selection that works — it’s portfolio structure.
@hetha_io
📉 Low-Liquidity Pairs: Hidden Execution Risks
In arbitrage, low liquidity means insufficient depth near the best price. The issue is not the percentage spread — the issue is whether the orderbook can absorb the intended volume.
A chain may appear profitable in calculation,
but limited near-price liquidity forces execution deeper into the book.
The weighted price deteriorates. Slippage increases. During recalculation, the chain becomes invalid. This is why in HETHA price difference alone is not enough. Execution feasibility depends on orderbook structure.
Arbitrage fails not because markets move fast - but because liquidity density is insufficient.
@hetha_io
In arbitrage, low liquidity means insufficient depth near the best price. The issue is not the percentage spread — the issue is whether the orderbook can absorb the intended volume.
A chain may appear profitable in calculation,
but limited near-price liquidity forces execution deeper into the book.
The weighted price deteriorates. Slippage increases. During recalculation, the chain becomes invalid. This is why in HETHA price difference alone is not enough. Execution feasibility depends on orderbook structure.
Arbitrage fails not because markets move fast - but because liquidity density is insufficient.
@hetha_io
🔄 Used Assets — the capital that actually works
In arbitrage, what matters is not the theoretical balance, but the capital that is actively used in chains.
In HETHA, this is reflected through Used Assets — funds utilized in chain calculations and executions within the selected period.
Turnover can be high, but the key metric is how efficiently the system utilizes available capital. A large balance alone does not create performance. What matters is the portion of funds that the market actually engages.
@hetha_io
In arbitrage, what matters is not the theoretical balance, but the capital that is actively used in chains.
In HETHA, this is reflected through Used Assets — funds utilized in chain calculations and executions within the selected period.
Turnover can be high, but the key metric is how efficiently the system utilizes available capital. A large balance alone does not create performance. What matters is the portion of funds that the market actually engages.
@hetha_io