TSPKey's System

TSPKey uses a proprietary group of timing models to determine our allocations. Our models are completely quantitative. This means they have objective, clear rules we follow 100% of the time.

TSPKey's timing models are based on a diverse set of market data and indicators such as

  • Trends in the overall market and the specific sectors of the TSP    Funds

  • Market breadth indicators like trading volume and the number of    advancing vs. declining stocks

  • Seasonal tendencies

  • Relationships among asset classes (large-cap stocks, small cap    stocks, bonds, commodities like crude oil, etc.)

  • The activities of large commercial traders

  • Interest rates and inflationary measures

    While each timing model is important, we've found that combining them into a composite system yields the best results. Our composite system assigns a degree of importance to each model. It then provides a weighted score based on the results.

    For our TSP analyses, we use two composite systems: one for the overall stock market and one for bonds. These two systems tell us which asset class offers the best risk/reward potential: stocks, bonds or risk-free (the G Fund).

    If our optimal asset class is stocks, we use a separate system to optimize a C, S and/or I Fund allocation. If it's bonds, we move to the F Fund. If neither stocks or bonds are favorable, we allocate 100% to the G Fund.

    While some of our ideas are original, we've also incorporated best-of-breed timing models from others. Everything we use has been thoroughly tested (for 60+ years in some cases) and has outperformed buy-and-hold investing. Many of our models originated from top-notch system developers, money managers and hedge funds. In some cases we use their original versions; at other times we use our own variations.

    Please note that our system development and "number crunching" require professional testing platforms and data feeds. If you paid for these, you'd spend much more than our annual subscription cost each year. You'd also need to spend countless hours doing your own testing and tweaking. So why re-invent the wheel?