The good, the bad, and latency: exploratory trading on Bybit and Binance
Bibliographic record
Abstract
We present the findings of a large-scale live trading experiment involving the placement of millions of market orders sent at a high frequency on two cryptocurrency exchanges, Bybit and Binance. We analyze the execution outcomes of these orders in comparison to the expected outcome based on the most recent snapshot of the Limit Order Book (LOB) at the time of order submission for two execution modes: one using market orders and the second using marketable limit orders aiming at the best price. Discrepancies between the actual and expected outcomes are due to intermittent LOB updates during a time span resulting from delays on the exchange, delays on the trader's end, or communication delays between the trader and the exchange. We show these discrepancies are strongly correlated with market factors such as volatility, latency, and LOB liquidity. Notably, we find a consistent disadvantage to the trader, pointing to an adverse selection effect for taker orders: profitable orders (as measured by short-term future PnL returns) tend to achieve worse-than-expected outcomes, while unprofitable orders typically achieve their expected (adverse) outcomes. In the case of market orders, this translates to a worsening of fill prices, while marketable limit orders suffer from a substantial probability of failing-to-fill-immediately. Quantitative researchers who fail to take these effects into account face the familiar litany of underperforming in a live trading environment relative to stellar backtests. To address this concern, we propose parsimonious models to estimate an order's probability of failing-to-fill-immediately (in case of a marketable limit order) and the worsening of its fill price (in case of a market order), allowing for greater accuracy when carrying out backtests and minimizing the discrepancy between backtest and realized live PnL.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.012 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".