A CNN Market-Making Idea for Predicting Price Moves and Hedgeability

Summary

The document proposes a convolutional neural network for market-making decisions and asks whether related research exists. Its suggested inputs combine current market depth, the trader’s previous position price and size, and a sequence of recent order-book snapshots. The proposed outputs are the time required to close both long and short positions and the resulting price change, framing inventory liquidation and price movement as prediction targets relevant to hedgeability.

This is an outline of a research idea, not a paper review or tested strategy. It gives no model architecture, labeling procedure, training data description, benchmark, or performance evidence, and it does not define precisely what it means to close both sides of a position. Those details would be needed to evaluate whether the proposed targets are learnable and useful for market making. The document’s contribution is to specify a combination of order-book history and inventory state as candidate inputs, alongside liquidation time and price change as candidate outputs.

Key ideas

• The proposed CNN uses market depth, recent order-book snapshots, and the trader’s prior inventory state.
• The suggested targets include position-closing time and price change.
• The idea connects short-horizon prediction with inventory management and hedgeability in market making.
• The document offers no cited studies, implementation details, or empirical results.

Tags

# Any papers on use of convolution neural network for predicting price and hedagability when market making


# Any papers on use of convolution neural network for predicting price and hedagability when market making












My idea:

CNN takes following input

```
Market Depth

My previous position price and position size when MM for current snapshot

Market book for past 10 to 30 snapshots
```

output:

```
Time taken to close both positions, long and short

Price change.
```

Are there any papers resembling my idea?

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.