Backtesting a Weighted Stock Scoring Model with Historical Data

Summary

The document describes a simple stock ranking model that combines market capitalization, liquidity, and value at risk. Each factor is divided into three intervals and assigned a score, then the scores are combined with weights of 25%, 25%, and 50%, respectively. The question is how to assess the model using historical data.

The response suggests using a rolling window when historical values for all three factors are available. This points to evaluating the scoring process over successive periods using information available at each point in time. The answer does not specify how to form portfolios from the scores, choose a holding period, measure performance, or account for trading costs. It is therefore a starting suggestion rather than a complete backtesting procedure. The document offers no empirical results or evidence that the proposed scoring weights or factor intervals are effective; those choices would need to be assessed separately, with care to avoid using future information.

Key ideas

• The model combines market capitalization, liquidity, and value at risk into a weighted score.
• Historical factor values are needed to evaluate the model through time.
• A rolling window is suggested as a basic way to conduct the backtest.
• The response does not define portfolio rules or performance measures.

Tags

# Backtesting a stock scoring model


# Backtesting a stock scoring model












I'm working on a simple stock scoring model consisiting of 3 factors:

1.market cap

2.liquidity of the stock

3.the value at risk

we defined 3 intervals for each factor and we assigned the intervals to scores (1 or 2 or 3)

the final rating is the following:

Score=25% marketcapscore +25% liquidityscore+ 50% varscore

So, how can i back test my scorig model ?

## Answer by Psi (score 0)

https://quant.stackexchange.com/a/49309

If you have historical values for market cap, liquidity and value of risk then you can backtest by simply using a rolling window.

If there is something in particular you are struggling with, consider editing your question.

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.