How does DPT differ from ChatGPT or Claude?

In the task. General language models produce plausible text — even when the underlying data is thin. DPT produces probabilities for market states and reports how robust these are. We use language models as a complement, to make results readable, not to generate them.

Which data goes into the model?

Market data, fundamental data, positioning and liquidity indicators, and structured text sources such as earnings calls and regulatory filings. No client data, no non-public information.

Is CaesarDPT a black box?

No. For every output the terminal shows the drivers, the historical comparison situations and the statistical robustness. The model weights are not individually interpretable — the derivation of a result certainly is.

The AI architecture behind CaesarDPT.

Understand how our proprietary AI model works and the architecture behind the signals.

What CaesarDPT is

Human and machine — in that order.

CaesarDPT is an in-house AI architecture specialised in recognising recurring patterns in financial time series. Where conventional models look at individual key figures, CaesarDPT processes entire price histories in context: price, momentum, volatility and market breadth, across markets and asset classes.

Every signal is dated, stored and verifiable. What CaesarDPT said yesterday cannot be rewritten today — our signal history is publicly documented accountability.

No model knows the market completely. CaesarDPT supplies the edge — the human carries the responsibility.

How CaesarDPT spots patterns before the market prices them in.

No magic box, no secret formula: a documented AI system whose signals you can check every day.

Signal

CaesarDPT analyses financial time series daily across markets, sectors and asset classes — on the basis of billions of data points from prices, volatility and market structure.

Grouping and combination

Recognised patterns are condensed into scores: which stocks show strength, which market regime applies, which weighting is appropriate — documented so it can be retraced.

Risk management

The regime governs the exposure — from defensive to overweight. Clear, transparent grounds for a decision. The final decision is always taken by a human.

See CaesarDPT in a live demo.

Let our technical team show you how it works in detail, and how reliable it has been historically.

Frequently asked questions

A language model learns which word follows another. A Deep Pattern Transformer learns the same principle on financial time series: which market constellation typically follows another. Instead of words it processes price, volume, fundamental and positioning data, and weights which past situations are relevant to today’s.