Track Record
The ContraQuant model is tracked live in real-time and always has been. Read the "Performance Integrity" section
below or see the
FAQs page to learn more.
This page summarises the track record. Please click on the yearly tabs on the left for in-depth summaries and
all report back issues. All reports were issued in real time. The chart on the
home page shows all entry and exit points.
The model has given mostly winning signals so far, albeit infrequent ones. Profits on individual trades have
ranged from 3.1% to 16.3% (scroll down for a full listing). The biggest and only loss on a trade was 20%.
As highlighted in the above chart, there have been consistent positive double-digit annual returns, except
for 2008 when the model was unfortunately not immune to the sharp market downturn.
A quick browse of randomly-selected live reports will show that the model's directional forecasts
and trades have been pretty consistent.
Again, with the exception of late 2008, monthly downside volatility has also been limited.
All historic trades are listed below. Levels always relate to the NASDAQ Composite. Please check the
individual real-time reports to verify the truth of our statements.
Our trading strategy is innately conservative and many traders could have used the model's accurate timing signals
to generate significantly higher returns. Read the
FAQs for further details.
The summary annual returns are as follows:
The latest cumulative return since the model's inception (May 2004) is +46.9% to January 2nd
2009 (trading an initial $1m starting capital), as we've reinvested profits into positions entered in later years.
The trading strategy was designed only to quantify the model's performance in real time, yet it has still beaten most
comparable benchmarks. Scroll down to read about "Performance Integrity".
The full list of trades is shown below. All trades show cost-average NASDAQ levels (most were scale-ins/scale-outs). Browse
the weekly reports on the yearly performance pages for details.
Please note that the figures below are sometimes affected by rounding e.g. the closing level of trade 1 is actually
1995.3 (since we scaled out of the trade on different dates so it is an average level).
This rounding factor might lead to minor discrepancies, however, in principle anyone can audit the trades below. Please
do so by downloading weekly real-time reports from the yearly "live performance" pages.
All profits are actual profits made in real-time on closed trades starting from a notional $1m. Some
of the trades below did not deploy our full capital. Again, for exact details please download the reports for the corresponding
period.
Trade 1: LONG
Entry on 24 May '04 @av. level of 1925
Exit on 30 June '04 and 6 July '04 @av. level of 1995
Profit: +3.7%
Trade 2: LONG
Entry on 6 August '04 and 16 August '04 @av. level of 1787
Exit on 12 November '04 @av. level of 2079
Profit: +16.3%
Trade 3: SHORT
Entry on 1 December '04, 2 December '04 and 3 December '04 @av. level of 2144
Exit on 2 February '05 @av. level of 2078
Profit: +3.1%
Trade 4: SHORT
Entry on 7 February '05, 14 February '05 and 15 February '05 @av. level of 2090
Exit on 18 April '05 @av. level of 1918
Profit: +8.2%
Trade 5: LONG
Entry on 26 September '05, 5 October '05, 6 October '05, 10 October '05 and 12 October
'05 @av. level of 2113
Exit on 19 December '05 @av. level of 2227
Profit: +5.4%
Trade 6: LONG
Entry on 22 May '06, 24 May '06, 6 June '06 @av. level of 2171
Exit on 28 Feb '07 @av. level of 2400
Profit: +10.5%
Trade 7: LONG
Entry on 5 Mar '07 @av. level of 2345
Exit on 30 Apr '07 @av. level of 2545
Profit: +8.5%
Trade 8: LONG
Entry on 14 Aug '07, 15 Aug '07 @av. level of 2484
Exit on 9 Oct '07 @av. level of 2799
Profit: +12.7%
Trade 9: LONG
Entry on 22 Jan '08, 6 Feb '08 @av. level of 2230
Exit on 20 May '08 @av. level of 2502
Profit: +12.2%
Trade 10: LONG
Entry on 14 and 15 July '08 @av. level of 2237
Exit on 18 Aug '08 @av. level of 2417
Trade 11: LONG
Entry on 15 and 22 September '08 @av. level of 2200
Exit on 7 Oct '08 @av. level of 1760
Please click through to
Contact and see if you are eligible to receive future real-time trading signals free of charge.
Although there has been only one losing trade since we began real-time weekly tracking in May 2004, we do not
expect to call the market right all the time. The model is based on quantifiable probability. Hence the loss
in 2008 is not entirely a shock.
It can however be said from the overall record of 10 closed winners and 1 loser since May
2004, that the accuracy of the model's market timing is unlikely to be a random fluke.
The model's performance generally has low correlation with other hedge fund strategies. This is confirmed
by clicking on the year by year reviews via the tabs on the top left.
See the
FAQs to learn how the model works in more depth and to walk through some of the real-time trades that have been executed.
Performance Integrity
Many forecasting models are built on back-testing of historic data. We reject such an approach. Markets move
in cycles and the historic context in which testing occurred may never recur.
Back-tested models are notorious for failing to hold up in real-world conditions.
Also, many of those who use back-testing to construct models revert to further back-testing when they
hit a rough patch. This constantly shifts the goalposts rather than coming up with a truly resilient approach.
By contrast, the ContraQuant approach has always rigorously tested the model in real market conditions by posting
all its trades in advance. We do this by sending weekly reports to our referees, including Jack Schwager.
The market has a habit of undoing the most well-intended assumptions and nothing can trump audited, real-time performance
sustained over a significant period of time.
As a result of the model's consistent performance, we have hardly needed to adjust its underlying parameters. This
is despite the continuous R&D we conduct on various sentiment indicators.
We have traded a notional $1m of capital in real time and verified and documented each trade meticulously through
our weekly reports.
We have been sending our weekly report with signals posted in real time to our independent referees weekly since May
2004. Click on the "Performance" tabs to browse past reports.
Each report is issued over the weekend based on the closing level on the previous Friday (or the last day the
market was open).
All trades are posted in advance using market and limit orders - even though this sometimes denies us the benefit
of optimal fills.
Our performance has been subject to the "messiness" of real-world conditions. For example, trades may in the past
have only been partially filled and we had no chance to rectify this until the next report.
We have never failed to file a weekend report on time (even if there was no trade pending). This has included public
holidays and vacation times.
The integrity of the real-time track record and live performance is of absolutely paramount importance.
Our referees can vouch for the timeliness and consistent receipt of the weekly reports and subsequent accuracy of
signals which were issued, always without benefit of hindsight.
Since our historical record of documentation is unbroken (you can check this yourself by downloading each of the historic
live reports, week by week) no trades could have been missed out from the above performance figures.
We are prepared to work with any potential partner who wishes to independently confirm the integrity
of the real-time track record.
