santa_fe institute

Embed Size (px)

Citation preview

  • 8/9/2019 santa_fe institute ...

    1/26

    Santa Fe InstituteMarch 29, 2005 1

    Gambling, Derivatives and

    Market Inefficiency

    Edward O. Thorp

  • 8/9/2019 santa_fe institute ...

    2/26

    Santa Fe InstituteMarch 29, 2005 2

    Gambling, Derivatives and Market Inefficiency

    ABSTRACTNobel prize-winning economists have claimedfor forty years that investors cannot to anysignificant extent beat the market through skill.

    This is the content of various forms of their efficient market hypothesis or EMH. But theyare wrong, as shown by the experiences of realworld investors, including my own, in the worldsof gambling and the securities markets. I

    explain the paradox and expect that you willimprove your ex ante expected stock marketresults no matter what type of investor you are.

  • 8/9/2019 santa_fe institute ...

    3/26

    Santa Fe InstituteMarch 29, 2005 3

    Efficient Market Hypothesis or EMH

    1960s

    Cootner (1964), Samuelson (1965), Fama(1965-70), Roberts (1967) and others

    Key consequence no way to beat themarket, i.e. excess risk adjusted expected

    return Still no consensus among financial

    economists, Lo (1997)

  • 8/9/2019 santa_fe institute ...

    4/26

    Santa Fe InstituteMarch 29, 2005 4

    False theories abound in the history ofscience

    EMH is false but useful Extensive evidence contradicts

    assumptions

    Anomalies Behavioral finance

  • 8/9/2019 santa_fe institute ...

    5/26

    Santa Fe InstituteMarch 29, 2005 5

    The Inefficient Market in Action 3Com Spins Off Palm Pilot

    Thursday, March 2, 2000 6% of palm Pilot

    was offered in an IPO Market valued palm pilot at $53.4 billion

    Therefore 3Coms 94% of palm Pilot was

    worth $50 billion Yet market valued 3Com at only $28

    billion

  • 8/9/2019 santa_fe institute ...

    6/26

    Santa Fe InstituteMarch 29, 2005 6

    Stub valued at negative $22 billion!

    Yet analysts valued stub between

    $5 billion and $8.5 billion

  • 8/9/2019 santa_fe institute ...

    7/26

    Santa Fe InstituteMarch 29, 2005 7

    Table 1 Price Disparity COMS and

    PALM Spin-off

    Trading Day COMS PALM PALM Premium

    1 Th 3-2-00 81.20 95.06 67 (83%)

    2 F 3-3-00 83.06 80.25 45 (55%)

    3 M 3-6-00 69.56 63.13 36 (51%)

    4 Tu 3-7-00 72.25 66.88 38 (53%)

  • 8/9/2019 santa_fe institute ...

    8/26

    Santa Fe InstituteMarch 29, 2005 8

    Short PALM and buy 3Com for a 100%profit

    PALM was difficult to borrow to sell short

    Any reasonable model of efficient markets

    investors always choose more moneyinstead of less money

    EMH: No investments A and B should

    exist where A dominates B Over a month for COMS/PALM disparity todecline to 10% or so

  • 8/9/2019 santa_fe institute ...

    9/26

    Santa Fe InstituteMarch 29, 2005 9

    Table 2 Continuing Disparity

    COMS and PALMTrading Day COMS PALM PALM Premium if COMS Stub = 15%

    of COMS

    5 W 3- 8-00 70.44 64.75 38%

    10 W 3-15-00 61.06 55.75 37%

    20 W 3-29-00 63.20 49.69 18%

    30 W 4-12-00 44.31 33.31 13%

    40 Th 4-27-00 38.88 27.06 4%

  • 8/9/2019 santa_fe institute ...

    10/26

    Santa Fe InstituteMarch 29, 2005 10

    The Ideal Efficient Market

    (You Cant Beat It)

    1. All information instantly available to all

    participants. COMS/PALM was frontpage news for weeks.

    2. Sufficiently many participants are

    financially rational, always prefer moremoney to less money, other things beingequal.

  • 8/9/2019 santa_fe institute ...

    11/26

    Santa Fe InstituteMarch 29, 2005 11

    3. Sufficiently many participants instantly

    evaluate all relevant information anddetermine the current fair price of everysecurity.

    4. New information: Prices immediatelygap to new fair price.

    Supporters claim this is good approximation

    (for large cap stocks in liquid markets).COM/PALMS example rebuts assumptions 2,

    3 and 4

  • 8/9/2019 santa_fe institute ...

    12/26

    Santa Fe InstituteMarch 29, 2005 12

    The Real Inefficient Market

    (Some of You Can Beat It)

    1. Information typically starts out known tolimited number, spreads to wider groupin stages. People who act earlier gain.Others dont.

    2. Financial rationality of the participants is

    limited.3. Participants typically have only some of

    the relevant information.

  • 8/9/2019 santa_fe institute ...

    13/26

    Santa Fe InstituteMarch 29, 2005 13

    Time and the willingness or ability toanalyze information varies amongindividuals.

    4. Buy and sell orders often spread overhours, days or months as academicliterature on Anomalies documents.

  • 8/9/2019 santa_fe institute ...

    14/26

    Santa Fe InstituteMarch 29, 2005 14

    Replacing EMH

    Positive alpha is a joint function of the

    market and the observer, further, eachobservers perceptions change over time.

    Two well-known historical examples

    illustrate this perfectly:

  • 8/9/2019 santa_fe institute ...

    15/26

    Santa Fe InstituteMarch 29, 2005 15

    1. Casino blackjack before and after itsmathematical analysis and the discoveryof card counting;

    2. Derivatives pricing before and after theBlack-Scholes formula and the ensuingrevolution in quantitative financial

    analysis.

  • 8/9/2019 santa_fe institute ...

    16/26

    Santa Fe InstituteMarch 29, 2005 16

    Participants in the market only gradually

    and sporadically used the new information. 45 years (1960-2005) for blackjack

    market to return to approximate

    efficiency. Derivatives markets, substantial

    inefficiencies have persisted for at least 35years (1967-2002) and perhaps continue.

    Dependence on the observer explains inpart why EMH debate continues.

  • 8/9/2019 santa_fe institute ...

    17/26

    Santa Fe InstituteMarch 29, 2005 17

    Investors

    1a. Investor believes the security (or bet) isproperly priced.

    1b. No view equals no case for mispricing,

    equivalent to his assuming it is properly priced.2. Ex ante, investor believes security not properlypriced, but is wrong.

    3. Ex ante, investor believes the security is properly

    priced and is correct. (Hedge COMS long andPALM short or swap from PALM to COMS.)

  • 8/9/2019 santa_fe institute ...

    18/26

  • 8/9/2019 santa_fe institute ...

    19/26

    Santa Fe InstituteMarch 29, 2005 19

    1. Before costs, the return on the averageactively-managed dollar will equal thereturn on the average passively-managed dollar; and

    2. After costs, the return on the averageactively-managed dollar will be less than

    the return on the average passively-managed dollar.

  • 8/9/2019 santa_fe institute ...

    20/26

    Santa Fe InstituteMarch 29, 2005 20

    These assertions hold foranytime period.Moreover, they depend onlyon the laws ofaddition, subtraction, multiplication and

    division. Nothing else is required. Active investors as a group incur an

    additional 2% per year in trading costs andmanagement fees.

    Diminished further through adverse taximpact of trading.

  • 8/9/2019 santa_fe institute ...

    21/26

    Santa Fe InstituteMarch 29, 2005 21

    Reason for persistence of EMH: Mostinvestors do not (and cannot)unequivocally identify positive (after cost)alpha.

    They are advised to act as though theEMH is true, and to invest either through

    low cost indexing or through buy andhold.

  • 8/9/2019 santa_fe institute ...

    22/26

    Santa Fe InstituteMarch 29, 2005 22

    Examples of Securities WithPositive Alpha

    In 1965 I began discovering these; from1967 until now I have invested in them.

    How have I done? Modest 1967 net worth is in 2005 6,680

    times as large, annual compound rate of

    26%. Net after costs, gifts, livingexpenses and taxes.

    Positive returns in all 38 years.

  • 8/9/2019 santa_fe institute ...

    23/26

    Santa Fe InstituteMarch 29, 2005 23

    Princeton Newport Partners

    November 1969 through December 1988compounded net, pretax at 15.1% nolosing years.

    Before general partners fee, 18.9%annually.

  • 8/9/2019 santa_fe institute ...

    24/26

    Santa Fe InstituteMarch 29, 2005

    24

    Statistical Arbitrage Operation

    August 1992 to October 2002 pretax 20%annually to limited partners.

    26% annually before general partnersfees.

  • 8/9/2019 santa_fe institute ...

    25/26

  • 8/9/2019 santa_fe institute ...

    26/26

    Santa Fe InstituteMarch 29, 2005

    26

    References

    7. Haugen, Robert A., 1999. The New Finance, The Case Against EfficientMarkets, Second Edition, Prentice Hall, New Jersey.

    8. Lo, Andrew W. (Edited by), Market Efficiency: Stock Market Behaviourin Theory and Practice, 1997.

    9. Roberts, H., 1967. Statistical versus Clinical Prediction of the StockMarket, unpublished manuscript [28]

    10. Samuelson, Paul A., 1965. Proof that Properly Anticipated PricesFluctuate Randomly, industrial Management Review, 6, 41-9. [185]

    11. Sewell, Martin, http://www.e-m-h.org/introduction.html

    12. Sharpe, W.F., The Arithmetic of Active Management, FinancialAnalysts Journal, January/February 1991, pp. 7-9.

    13. Stein, David M., Active and Passive Arguments: In Search of an

    Optimal Investment Experience, Journal of Wealth Management,Winter2003, pp. 39-46.

    14. The New York Times, March3, 2000, page A1, Offspring UpstagesParent in Palm Inc.s Initial Trading.