Momentum exists - the "Frog in the Pan&quot

From one of the deepest research firms and strongest proponents of modern quantitative approaches - Alpha Architect - https://alphaarchitect.com/2022/02/01/what-explains-the-mome…

We walked away from the question and determined the following:

We will never really know why the momentum factor actually exists. But we know that is does exist and it is arguably stronger than value.
The arguments as to why momentum exists can be boiled down into two primary camps:
momentum is an underreaction phenomenon
momentum is an overreaction phenomenon.
Our most informed guess is that momentum is driven by underreaction and this is best captured via the so-called ‘From-in-the-Pan‘ algorithm, which we use in our own investment processes — i.e., Quantitative Momentum. https://alphaarchitect.com/2015/12/01/quantitative-momentum-…

Oversimplified, frog in the pan implies gradual sustained price changes, not sudden.
https://alphaarchitect.com/2015/11/23/frog-in-the-pan-identi…
The conclusions are clear: a more sophisticated momentum strategy that focuses on the path-dependency of momentum generates a much stronger momentum effect.

Worth a deep read and further discussion, IMO.

5 Likes

Frog in Pan (FIP) sounds similar to what MTUM does: combine high momentum and low volatility. The Alpha Architect ETF QMOM 5-year returns are lower than some other momentum ETFs.

                                              Total Return
                Name                  Ticker      2017      2018  2019  2020  2021  2022YTD  5-year
    Invesco S&P MidCap Momentum        XMMO        37         6    37    29    17     -6       21
        Invesco S&P 500 GARP           SPGP        36         2    39    16    36     -8       21
      Invesco S&P 500 Momentum         SPMO        28        -1    26    28    23     -6       18
  iShares MSCI USA Momentum Factor     MTUM        38        -2    27    30    13     -10      17
Invesco S&P SmallCap Value with Momt   XSVM         4       -12    30     5    56     -2       14
Alpha Architect US Quantitative Momt   QMOM        16       -12    28    62    -4     -4       13

https://gtr1.net/2013/?~kindaLikeMTUM2_20210424_borisnand:h1…
https://gtr1.net/2013/?~Nas100Momentum_20191030_rgearyiii:h2…

10 Likes

Frog in Pan (FIP) sounds similar to what MTUM does: combine high momentum and low volatility.

Which sounds like the RSS family of screens. What ever happened to them? They seem to have been dropped from the listings.

DB2

2 Likes

MTUM has been a solid performer for a few years. But my point in posting it really was, the summary of the two articles is that sustained smooth incremental price increases momentum


                   -
            -
      -     
- 

is better than spiked momentum

                    |
                    |
          ----------
----------

and therefore it might be better to develop, or redevelop, screens that rely on sustained momentum, and avoid spiked momentum - and maybe focus on the mid-capish world.

5 Likes

Adding volatility criteria to a screen often reduces both CAGR and GSD. It might be useful for screens with very high GSD. Adding a final step that drops the stock with the highest volatility sometimes increases CAGR. For example, {TechGrowers2_volatility} uses:

step5: [Total Return % over 126 days] Top 6
step6: [TR1-Volatility over 126 periods] Bottom 5

Screen                       CAGR  GSD  MDD  Sharpe
TechGrowers_20200326_rdutt    46   59   -58   1.02
TechGrowers2_20200326_rdutt   53   58   -54   1.14
TechGrowers2_volatility       59   53   -50   1.29

from 20170210 to 20220211, depth 5, friction 0.4
https://gtr1.net/2013/?~TechGrowers_20200326_rdutt:h21f0.4::…
https://gtr1.net/2013/?~TechGrowers2_20200326_rdutt:h21f0.4:…
https://gtr1.net/2013/?~TechGrowers2_volatility:h21f0.4::sty…

9 Likes

These are today’s picks for {Tech_Growers2}. Comparing CARG to BILL in a price chart gives an idea of the volatility measure.
https://stockcharts.com/freecharts/perf.php?CARG,BILL&n=…

Tech_Growers2
Ticker Symbol        Company Name       styp.a  dspo:1  aprc   [MktCap]   trbc7.s  salesgq5q1d.s  trp:1,126  vol:1,126,1
    PANW         PALO ALTO NETWORKS       11     2407    510     50,334   5720103        32           36         0.43
     APP         APPLOVIN CORP CL A       11      210     70     15,919   5720101        90           25         0.69
    DDOG           DATADOG INC CL A       11      605    167     43,055   5720102        75           24         0.57
    INST             INSTRUCTURE          11      142     23     3,268    5720103        31           16         0.52
 **CARG          CARGURUS INC CL A       11     1091     35     3,522    5720103        51           16         0.41**
 **BILL               BILL COM           11      546    237     24,284   5720102       156           14         0.84**
     MDB           MONGODB INC CL A       11     1086    427     28,515   5720102        51           13         0.69
     KBR               KBR INC            11     3835     44     6,206    5720101        34           13         0.27
      ZS             ZSCALER INC          11      985    273     38,242   5720102        62           11         0.54
    ZNGA           ZYNGA INC CL A         11     2555     9      10,356   5720103        40           11         0.61
    DOCN             DIGITALOCEAN         11      225     59     6,460    5720101        37           11         0.82
    ABNB           AIRBNB INC CL A        11      295    167     57,757   5720103        67           9          0.51
    CFLT         CONFLUENT INC CL A       11      161     57     3,976    5720102        67           6          0.90
     IAS         INTEGRAL AD SCIENCE      11      157     19     2,902    5720102        32           4          0.54
    JAMF          JAMF HOLDING CORP       11      394     34     4,051    5720102        36           3          0.58
    SNOW         SNOWFLAKE INC CL A       11      355    294     89,957   5720101       110           1          0.56
    FTNT             FORTINET INC         11     3079    310     50,729   5720101        33           0          0.46
     NOW           SERVICENOW INC         11     2421    584    116,744   5720102        31           -1         0.42
    GDYN            GRID DYNAMICS         11      827     22     1,448    5720101       120           -1         0.71
     TYL         TYLER TECHNOLOGIES       11     7994    468     19,163   5720102        61           -2         0.30
    TTGT           TECHTARGET INC         11     3712     76     2,198    5720101        93           -2         0.55
    GOOGL         ALPHABET INC CL A       11     4402   2,686  1,654,457  5720103        32           -3         0.27
      EA           ELECTRONIC ARTS        11     8163    134     37,712   5720102        59           -3         0.28
    TEAM              ATLASSIAN           12     1554    322     45,418   5720102        34           -5         0.49
     TTD       THE TRADE DESK INC CL A    11     1358     76     33,273   5720102        39           -6         0.75
     ZEN             ZENDESK INC          11     1951    116     14,138   5720102        32           -6         0.45
    GLOB             GLOBANT S A          12     1907    249     10,353   5720101        65           -8         0.50
      S          SENTINELONE INC CL A     11      157     45     1,879    5720102       128          -10         0.74
    APPS         DIGITAL TURBINE INC      11     2184     49     4,747    5720102       338          -11         0.72
    SMAR         SMARTSHEET INC CL A      11      956     62     7,842    5720102        46          -11         0.57
4 Likes

Adding volatility criteria to a screen often reduces both CAGR and GSD. It might be useful for screens with very high GSD.

If you are compounding, then reducing high GSD often increases CAGR.

DB2

4 Likes

If you are compounding, then reducing high GSD often increases CAGR.

Not just often, that’s the general rule for individual securities.
Low volatility stocks have always had higher average returns than high volatility stocks over time.
CAPM has always been hilariously backwards comparing equities to equities.

Comparing equity portfolios to equity portfolios is a bit more nuanced.
Sometimes low volatility portfolios perform better, sometimes worse.

Jim

9 Likes

One step in the paper is eliminating stocks with recent high daily returns.

The authors use the MAX measure, defined as the average of the highest five daily stock returns over the past month.

Are there any suggestions in how to calculate a Field in GTR1 for getting the average of the highest 5 daily returns in the month for a stock?

Craig

1 Like

The authors use the MAX measure, defined as the average of the highest five daily stock returns over the past month.

In my 6/3 options strategy I first rank the stocks by total five day return (most recent five days) and remove the top 10% of stocks from consideration.

Elan

7 Likes

In my 6/3 options strategy I first rank the stocks by total five day return (most recent five days) and remove the top 10% of stocks from consideration.

Would the inverse be applicable to put candidates?

DB2

Would the inverse be applicable to put candidates?

If you believe in the value support of the firm in question, yes.
An ideal pick is one that just dropped in price. Cheaper entry, and time value on options will have jumped, so you win on both fronts.
But that approach is unlikely to work well to get a large number of picks on any given day.

And would occasionally present a firm that just had a price drop that was justified, not just random noise.
e.g., a big fraud.

Personally I like cash backed put writing a lot, but my only edge is having a confident* opinion of what constitutes a good price for a given stock.
So I don’t think it is a great fit for quant selection criteria.
I’ve written a LOT of put contracts (over 100k of them), but only ever against only about 65 tickers total.

  • Note, “confident” doesn’t mean “correct”!

Jim

5 Likes

One step in the paper is eliminating stocks with recent high daily returns. The authors use the MAX measure, defined as the average of the highest five daily stock returns over the past month. Are there any suggestions in how to calculate a Field in GTR1 for getting the average of the highest 5 daily returns in the month for a stock?

I don’t know how to easily get the highest 5 daily returns. Can you post a link to the paper that uses that definition? MAX is a terrible name for a variable, as it is difficult to search for. The standard academic MAX is one day similar to maxret:

maxret: max(trp(1,1),trp(2,1),trp(3,1),trp(4,1),trp(5,1),trp(6,1),trp(7,1),trp(8,1),trp(9,1),trp(10,1),trp(11,1),trp(12,1),trp(13,1),trp(14,1),trp(15,1),trp(16,1),trp(17,1),trp(18,1),trp(19,1),trp(20,1),trp(21,1))

— links —
Frog in the Pan: Continuous Information and Momentum, 22 Dec 2013
“To account for extreme returns, we include the maximum daily return over the prior month (MAX)”
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2370931

Maxing Out: Stocks as Lotteries and the Cross-Section of Expected Returns, 15 Dec 2021
“a negative and significant relation between the maximum daily return over the past one month (MAX) and expected stock returns”
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1366204

3 Likes

For the largest 1-day percent gain over the last 21 market days:


maxrtn: tsmax(trp(1,1),0,21,1,step0)

Eric

3 Likes

Can you post a link to the paper that uses that definition

https://alphaarchitect.com/2014/06/09/betting-beta-demand-lo…

See Alternative Alpha section.

This paper documents that the betting against beta (BAB) strategy is mainly driven by investors demand for lottery-like stocks. The authors use the MAX measure, defined as the average of the highest five daily stock returns over the past month, to proxy for the lottery-demand of a stock. Figure 1 shows a heat map of the stocks that fall into portfolios measured by Beta and the MAX (lottery demand) measures:

The link in the paragraph for “MAX measure” takes you to a paper that has MAX more in line with your definition.

Regards

Craig

1 Like