Factor Ranking Portfolios' Backtesting

Factor Ranking Portfolios' Backtesting

 

To construct a portfolio with the best strategies implemented, we decide to take other great strategies as reference and build a strategy that will be proved effective through the method of back-test.

 

We started by examining the four basic models in the ranking system, which are the growth model, momentum model, quality model and value model, because we believe these models cover the basic but essential part of how a strategy forms. To analyze these models, we run the back-test and see how the performance was of each of the models and analyze how these models are affected by different selecting factors with the reverse engineering technique. We can see the past performance of each of the basic strategies as the below graph.

Quality

Growth

Momentum

Value

 

From the graph we could see even though the models are somewhat predictive, their performance aren’t very significant and can be inconsistent during time to time. After acknowledging this fact, we try to parse out the effective parts of each of the models, and by combining those parts we expect to see a better model with better predicting power and more consistency.

Using the reverse engineering technique, we take the top contributing factors from the basic models and put the factors into our new model.

Momentum Strategy

price_change_120; price_change_140; price_change_160;

price_change_180; up-down_ratio_20

 
 

Quality Strategy

ROE5%5YAvg; ROI%5YAvg; CurRatioQ; DbtTot2CapQ;

IntCovTTM;

 

Value Strategy

PEGLT; Pr2CashFITTM; ProjPECurFY; PEExcXorTTM;

Pr2BookQ

 

Growth Strategy

EPS5YCGr%; EPS_Acceleration_long_term; Sales5YCGr%;

Sales_Acceleration_recent; Sales_Acceleration_Longer_term

 

 

 After putting all these factors into our model, we run again the back-test and see how our new model performs. The performance is as the graph below. We could see the performance has improved dramatically with the performance of our portfolio constructed with top selected stocks outperforms S&P 500 significantly. However, from the chart we could still discover some inconsistencies and we wonder whether we could keep improving our model by leveraging the reverse engineering technique. Thus, we run again the technique and tune the weight of each factors through considering the reverse engineering result. The result is as the following graph at the right side. While the performance of the portfolio with the best stocks is similar to the previous model, we could see that the model delivers a more consistent performance with different level of stocks.

In conclusion, we believe we have constructed a good model with solid and consistent performance.

   

 

Final Selecting Factors and Explanation

Strategy

Variable

Weights

Formula

Explanation

Momentum

120+

5%

close(0) / close(120)

We use this variable to detect the influence from past stock price.

140+

5%

close(0) / close(140)

We use this variable to detect the influence from past stock price.

160+

5%

close(0) / close(160)

We use this variable to detect the influence from past stock price.

180+

5%

close(0) / close(180)

We use this variable to detect the influence from past stock price.

up-down ratio 20+

5%

UpDownRatio(20,0)

To help evaluate the relationship between the volumes of advancing and declining issues on an exchange.

Quality

ROE%5YAvg

5%

Return on Average Common Equity, 5 Year Average (%)

Use 5 years of past ROE to see how the company is performing.

ROI%5YAvg

5%

Return on Investment, 5 Year Average (%)

Use 5 years of past ROI to see how the company is performing.

CurRatioQ

5%

The ratio of total current assets for the period divided by total current liabilities.

To evaluate the short-term liquidity of the company.

DbtTot2CapQ

5%

This ratio is total debt divided by total capital.

To evaluate the leverage level of the company.

IntCovTTM

5%

Times Interest Earned

To evaluate the ability to pay back interest by the company.

Growth

EPS5YCGr%

5%

Earnings Per Share, 5 Year Growth Rate (%)

To evaluate the company growth by EPS growth rate.

EPS Acceleration - longer termƒ(x)

5%

( EPS%ChgTTM- EPS5YCGr%)/abs( EPS5YCGr%)

To evaluate the company growth by how EPS grow in the long-run.

Sales Acceleration - longer termƒ(x)

5%

( Sales%ChgTTM- Sales5YCGr%)/abs( Sales5YCGr%)

To evaluate the company growth by how sales grow in the long-run.

Sales5YCGr%

5%

Sales, 5 Year Growth Rate (%)

To evaluate the company growth by the 5 years sales growth.

Sales Acceleration - recentƒ(x)

5%

(Sales%ChgPYQ- Sales%ChgTTM)/abs( Sales%ChgTTM)

To evaluate the company growth by how sales grow recently.

Value

PEGLT

5%

Projected Price/Earnings to Long Term Growth Rate

To evaluate the project price divided by earnings.

Pr2FrCashFlTTM

5%

Price To Free Cash Flow Per Share Ratio, TTM

To evaluate the ratio of stock price to free cash flow per sharpe ratio

ProjPECurFY

5%

Current Fiscal Year Projected P/E Ratio ($)

To evaluate the projected P-E ratio.

PEExclXorTTM

5%

Price To Earnings Ratio, Excluding Extraordinary Items, TTM

To evaluate the stock compared to the earnings.

Pr2BookQ

5%

Price to Book Ratio, Quarterly

To evaluate the price to book ratio in order to understand the ratio of stock price and book value.