Month: September 2018

Past results do not guarantee future returns.

After reviewing several cycles, the 412 algorithm’s stocks follow an interesting pattern.

The specific 412 algorithm was made to increase on market returns, but it does not claim the ability to never drop in value, in theory or in design. It simply outperforms the market.

That is, the system performs about as you would expect, stocks it picks will typically drop in value when the entire market does, and will typically rise when the market does. In the event of a longer market downturn, the stocks value sometimes drop more than the market. But it more than makes up for this by far exceeding the market returns during a market recovery. Those are the specific attributes.

This is actually common in the world of value investing. It is also why the system works best when it invests in a wide field of many stocks. Because stocks of good value are valuable, they are not immune to market downturns, but will recover their stock prices and more afterwards, weathering storms. This means that, during a downturn, it is highly inadvisable to sell.

In fact it is wise to increase investments, as there are more value opportunities when others are selling stocks heedlessly!

Warren Buffett said it well: “Be fearful when others are greedy and greedy when others are fearful.”

Want a scientifically proven algorithm that finds the best investments?

To the best of my knowledge and abilities, the 412 system has been scientifically proven over the last 5 years.

There was a hypothesis, that high value stocks can be found through an algorithm, which would beat the market. There was over 3 dozen tests on backtests, forward tests, and side-by-side tests. Each test could fail, that’s part of the scientific process. Each test has shown a system that convincingly outperforms the market, showing the algorithm works as planned.

Backtests are when you run the algorithm on past data and check the output. This was done on dozens of spreadsheets using historical data generated by reading the xml of past balance statements and income statements.
Their results were distilled into a standard format and given a timestamp. One day every three months, stocks were picked based on data that would have been available on that day, to sort all stocks in a market by a certain criteria and pick those with the best criteria. Then the output was compared over 6-7 years against the actual market performance. Would these investments have returned better results, based on buying them every three months given their market price? The answer was a resounding no for most systems, but a few systems, like the 412 system, did consistently better than the market.

Caution demands you do side-by-side and real investments as well. So my own money, from my IRA, was invested using the same methods. At the same time, portfolios were created in google portfolio and finviz portfolio to check that picking stocks in real time at slightly different points would also show robust earnings above the market rate.

All three tests confirmed that the 412 system is robust and beats the market over the time period consistently.

My actual investments, possibly due to trading fees, bid spreads, and slow reaction time, did the worst of the three, but still beat the market by quite a bit. This lowest result is what my website is claiming: 21% average per year.

In addition, “filters” were scientifically researched, looking for ways to exclude stocks from the sort, if they were under SEC investigation, or had inventories growing and revenues shrinking, or if they showed other signs of fooling around with their numbers (since all decisions are based on the numbers, it only makes sense to then filter out those who put out incorrect numbers on purpose.) Some manual and automatic filters also improved results and consistency, they are part of system along with the original 412 algorithm.

The system is based on an independently verified “Low PE” strategy.
That strategy is to sort all stocks by PE, and pick the best ones. This was scientifically proven and verified by Fama and French in the 1960s. My system, being new, has not itself been independently verified yet.

As you can see, the scientific method is at the root of my testing. My own money is in the system, and has done well. It is scientifically proven, since it is based on a scientifically proven method (PE sorting), extends this, improves it with filtering. Neural nets and genetic programming found the algorithm in the first place, which was done on test and training sets.

This form of algorithmic discovery is also a miniature, automated, version of the scientific method.
It was tested in three different ways, each of which could have failed, but each test succeeded: real results, side-by-side, and backtests.

All of this is in well known brokerages. Anybody who requires that level of proof can simply see my results in real time on these systems as needed, to see for themselves that the results are as claimed. Also, portfolios and side-by-side stock picks from the present and past are available on request. Please set up an appointment with me and ask to see them.

Please note: Past results do not guarantee future returns.

Want a scientifically proven algorithm that finds the best investments?