There are a lot of asset allocation software choices for the professional – Morningstar’s Encorr, MSCI Barra – but not many for the individual. Very few are web based, and most are very expensive.
What are some good sites out there besides AssetPlay?
(I’ll update this post with $$ and links as I get to it.)
Zephyr AllocationAdvisor ($5,000) Mean-variance optimization (MVO), Black-Litterman, etc.
Pertrac Analytics ($5,500 – $8,300 /year)
SmartFolio ($600 – $6,000) Multi-period MVO, Black-Litterman, etc.
Morningstar’s Encorr ($10,000 / year)
Windham Portfolio Advisor ($12,000/year), Financial Planner ($6,000/year), Planner Express ($1,800/year)
Efficient Solutions ($70-$170 one time) Single and multi-period MVO, Monte Carlo.
DynaPorte ($20,000 /year) Macro factor based tactical asset allocation.
FastTrack ($400-$700/year) Poor man’s version of FastBreak.
FastBreak ($500, Pro $1,800) Mutual fund and stock trading system to test and trade strategies using fund rotation. Pro version features genetic algos.
Quantext ($100 – $180/year) Portfolio simulation.
Hi Mebane – I have seen it mentioned that your timing model is very sensitive to turn-of-the-month effect, meaning that much of the model's performance is due to your picking the last day of the month as model's trade-date.
Stated differently, this would mean that if one were to choose the 15th of each month – then the performance would be much lower.
I've read through your paper and have seen some rigorous testing of the model's sensitivity. So I'm wondering if you had tested this as well, and if you have any thoughts on the matter.
Thanks in advance, keep up the good work !
Cori – the choice of what day you update the model should have zero effect (in the long run).
Matlab.
Portfolio123.com
doesn't do asset classes, does it? if it does it only goes back to '00 if i
remember correctly. . .
Doing asset allocation backtesting with Investor's FastTrack is kinda like eating noodles with a socket wrench.
Wrong tool for the job. But a nice program for tracking mutual funds, ETFs, CEFs, and indices. I have been using it for about ten years.
Finally found where I had read this:
http://www.bogleheads.org/forum/viewtopic.php?p...
Take a look, long discussion in that link:
“I'm simulating Faber's once-a-month frequency trading system… it shows the volatility of the Faber type timing system over all possible “day-of-the-month” sets. Faber only backtested checking on the last calendar day of the month. Suppose he had picked some other day, would that make his system backtest differently? Turns out it would and there are big variations between lucky day picks and unlucky ones.”
Market Riders? not sure if you can back test, my trail subscription expired…
Good point – although the new ETF tools are pretty slick and could be used for asset classes.
The system works in virtually every market I have tested (not just US Stocks), and has been verified across time frames, parameters, and dates (independently, by dozens of pros and individuals). The basis for day of the month argument is not logical – over the long run. Obviously there will be short term variation, but it will be a wash in the long run.
I am looking for option strategy backtesting systems/software/tools/whatever…
Can anybody help?
Rmetrics is becoming a very nice open-source solution.
I've not seen any software for options backtesting – I'm not sure it exists because of the complexity and finding good data. I hope someone else on this thread will prove me wrong.
[...] I was flipping though Mike Carr’s book again the other day, and decided to download the monthly data for the Fidelity Select Sector funds from Yahoo. (Part of the reason was due to my frustration with the state of asset class testing software that I did a post on the other day). [...]
Damian: Data are one thing, but it would suffice for the beginning to have a simple model with some kind of random walks (lognormally distributed) with the Black Scholes formula – but with the possibility of systematic and dynamic trading and see the outcome in form of some density kernel/histograms.
You can add complexity later (like different distributions and so on).
My guess is that you'd have to go with something like R and the related tools (see fosstrading.com for more info) – because I don't know of software that does very simple testing, let alone more complex testing. Maybe tradestation does it? I suppose I could do it in Amibroker if I had the data. But the intraday spreads on options make it very difficult to test imho.
One of the best packages when it comes to asset allocation software for professionals is from Alternativesoft (http://alternativesoft.com ). 15k – 40k /year – depending on which modules you want/need .
The only options software/platforms with archived options price/volatility data I know of are:
Platinum (end-of-day data, web-based, subscription with Optionetics.com)
OptionVue (hourly or half-hourly data, software subscription)
Thinkorswim's thinkback tab (end-of-day, broker platform, need acct, new/primitive but update coming)
They provide options data for approx the last 5 to 10 years.
All of them are not really “backtesting” – they can be used to “backtrade” strategies. Ie, you cannot set parameters and run automatic backtesting over the data and get a table of results/equity curves as you can with stock backtesting. They work by setting the calendar back to a previous date and then you trade your way forward in time. They are good at allowing you to practice options trading skills over a range of market conditions.
Quantext Portfolio Planner
http://www.quantext.com/
$99/year
MPI Stylus
http://www.markovprocesses.com/products/mpistyl...
Hi to Mebane and to the readers of this thread,
I would just like to comment on the usage of such software and make you aware of a few flaws which can be met when using such tools.
Most of these software are made of two parts : an historical simulation, which is fine, and a forecasting module, which is most of time built on a theoretical model called “Monte Carlo” or “Brownian motion” simulation. This model makes the assumption that the financial returns follow a gaussian distribution, which has been proved wrong by recent theoretical works made and published by Mandelbrot (The misbehavior of markets) and Taleb (The Black Swan).
Using such software could give you the illusion that you are relying on very sound theoretical grounds, which are commonly used in finance, whereas the most recent research in mathematics have proven such theoretical grounds to be weak and not reliable for financial markets.
In the same order of idea, the calculation of historical correlation between asset classes, does not protect you from any change in such correlation in the future ; for instance, last october/november, we have seen correlation between asset classes converging to 1, and therefore portfolio diversification to not operate any more, which has led diversified portfolio to behave as if they had only one asset class.
These two flaws must not prevent your from using such software ; but you need to keep in mind that the strong technical background they are built on is not free from any defect and surely not take their future simulations for granted.
All the best,
Pierre
You may want to look at this public-domain effort by “Simba” (a contributor to the Bogleheads forums):
http://www.bogleheads.org/forum/viewtopic.php?t...
You can download from this posting a free Excel workbook.
Here are a number of things this free Excel model does:
- Compose a portfolio by assigning percentage share to multiple asset classes, choose from 27 classes
- Set starting year, ending year, initial investment, minimum acceptable return (for Sortino ratio)
- Compute portfolio growth, CAGR, Std Dev, Sharpe, Sortino (all with our without annual rebalancing)
- Chart annual portfolio growth trajectory (both nominal and real returns)
- Chart annual maximum drawdowns
- Easily compare performance parameters for up to 5 different portfolios
- Also includes performance parameters of a number of ‘lazy portfolios’, including some recommended by well-known “gurus”
Annual performance data for the asset classes used in this model goes back to 1972.
That is what I did. Used R and downloaded info from yahoo.
Logical Information Machines http://www.lim.com produced a cool data set that chains options together for backtesting based on how far out of the money that they are….so the X stike call option out of the money will be tracked and conjoined with the next contract after it expires. If you get the data from LIM, you can integrate it into portfolio optimizations with other options or stocks etc. using Gsphere from Gravity Investments. http://www.gsphere.com
Thanks for the list Mebane.
I've been thinking of using one of the lower priced vendors you had listed … SmartFolio — however been hesitant to do so first given my quandary with the whole concept of MVO because of the idea of that to get around from getting an end result where in which an unusual over allocation is put into one asset you have to put lower and upper limit thus taking away the whole idea of using a quant method to determine the allocation % — not too mention that I'm also not too verse on the BL model which they utilize.
So I'm curious if anyone else has had any experience with SmartFolio and if you have any insight you can share with a novice like me.
Hi gvm,
I have been a user and a big fan of Smartfolio for some years now. I find Boris, the person who is in charge of the software, very helpful in supporting his product.
The product has its limitations, but it is now reasonably stable and useful.
Not understanding the full mathematical theory behind it is not necessary to make an interesting use of this tool. You need to be careful though to the following :
- in case you are investing via ETFs or other mutual funds, the performance of the said ETFs is to be compared to the appreciation of cash. At present time, cash returns nothing, but it has not always been the case and it might change in future in inflation comes back. To get a proper asset allocation optimization, you need to have a time series with the actual return of cash, and to define it inside Smartfolio as the risk free asset.
- in case you are investing through futures contract, you can avoid this problem because most of the cash remains yours and you can put it in money market funds which gives you the appropriate return for your cash.
This is obviously independent of the software you are using.
Hope this helps.
Pierre
Hi gvm,
I have been a user and a big fan of Smartfolio for some years now. I find Boris, the person who is in charge of the software, very helpful in supporting his product.
The product has its limitations, but it is now reasonably stable and useful.
Not understanding the full mathematical theory behind it is not necessary to make an interesting use of this tool. You need to be careful though to the following :
- in case you are investing via ETFs or other mutual funds, the performance of the said ETFs is to be compared to the appreciation of cash. At present time, cash returns nothing, but it has not always been the case and it might change in future in inflation comes back. To get a proper asset allocation optimization, you need to have a time series with the actual return of cash, and to define it inside Smartfolio as the risk free asset.
- in case you are investing through futures contract, you can avoid this problem because most of the cash remains yours and you can put it in money market funds which gives you the appropriate return for your cash.
This is obviously independent of the software you are using.
Hope this helps.
Pierre