How GasPredictor.com's Atlanta Gas Predictor Newsletter Saves You Money

On this page:

• The Bottom Line
• Description of the test scenario
• Highlights of the test
• Table showing day-to-day events

The Bottom Line of the Test

Our subscriber paid $111.95 for gas since 2/17/2009. Our control driver paid $102.18. Our subscriber is ending the test with his tank nearly full, and the control driver's tank is less than half full. The extra 6 gallons in the subscriber's tank is worth $11.39. Subtracting that from the amount he spent for gas means that he spent $1.62 less than the control driver did. Extrapolating this savings over a full year (252 business days) yields an annual savings of $12.37.

Following our recommendations would save you $12.37 per year, or $7.78 (*) more than the cost of a one-year subscription to the Gas Predictor Newsletter.

* Savings based on the current regular subscription price of $4.59. Save even more in April, at $3.99.

And as a bonus, you get to smirk when you see the price go up and you knew it was going to go up. No extra charge.

Back to top  Back to 100th Prediction Summary

Description of the Test Scenario

This test compares two hypothetical drivers, the "subscriber" and the "control driver." Each has a car with a 14 gallon tank, each uses exactly two gallons of gasoline every business day, and each has just filled his tank the evening of February 16, 2009.

Our "control driver" does not subscribe to the Gas Predictor Newsletter. He simply leaves his gasoline budget to chance. He fills his tank whenever he gets home from work with only two gallons in the tank. Thus, he buys 12 gallons of gas every sixth business day, without regard to the cost. This represents the way most people buy gas, and shows how much a hypothetical driver would pay for gas without the benefit of our Gas Preditor.

On February 17, our "subscriber" begins receiving his Atlanta Gas Predictor Newsletter by e-mail. Each night, he tries to follow our recommendations, even though he can not always do so. If we recommend buying gas that day, he fills his tank, even if he just filled it the night before. If we recommend holding off and buying gas the next day, he does not buy any gas unless his tank is absolutely empty. If his tank is empty, he buys four gallons of gas, to get him through the next couple of days, and yet leave enough room in his tank that he can take advantage of any buying opportunity that our newsletter recommends.

We recognize, of course, that this test is not quite realistic in several ways. For one thing, nobody uses exactly two gallons of gas every single day. And nobody drives their car only to work, not driving at all on non-business days. Nobody has off every day that the NYMEX commodity exchange is closed. Nobody buys gas by the gallon, instead rounding up to the next dollar or otherwise easy amount of money to enter in their checkbook (or to make change, if they're paying in cash). And even though our predictions are perfectly wonderful, we're sure none of our subscribers would bother to buy just two gallons of gas because we recommended it. (No, even we don't do that.) However, we had to come up with fixed, simple rules in order to make the test a fair comparison between using our predictions and not using them.

For this test of the Atlanta Gas Predictor Newsletter, the price of gas used is the second-lowest price for unleaded regular gasoline in the Atlanta area at about 4:00 PM Eastern.

Back to top  Back to 100th Prediction Summary

Highlights of the Test

• A coincidence of timing: Three out of five times the control driver fills up, every six business days, happens to be a day when we recommended buying gas. He just got extraordinarily lucky. If not for this turn of luck, our hypothetical subscriber might have saved even more. We could have rigged the test, perhaps by having our control driver start with half a tank instead of a full tank, but we just played it honest.
• There were five large single-day price increases during this period: 4 cents on 2/25, 5 cents on 2/27, 7 cents on 3/5, 8 cents on 3/24, and 6 cents on 3/27. By coincidence, our control driver filled up within a day or two before three of these days, including the big one on 3/24. Again, this dumb luck saved our control driver a good bit of money, reducing our hypothetical subscriber's relative savings.
• February 25: This was the one and only time our predictions have been wrong. If we had correctly called the next day's price increases, our hypothtical subscriber would have filled up this day instead of the next, and he would have saved four cents per gallon.
• March 12: A large single-day price decrease, 5 cents per gallon. By another coincidence, our control driver fills up this day, accidentally reaping the benefit. (Maybe we should have rigged the test after all?)
• Just out of curiosity, and to show you how honest we are, we repeated the test scenario with different starting conditions. If we had started with our control driver at 2 gallons in his tank, buying his first 12 gallons on day one (2/17), our hypothetical subscriber would have had an annual savings rate of $148.53, over 18 times the annual subscription to the daily Gas Predictor Newsletter. The coincidences in this test are indeed a major factor, and we believe that our results will improve dramatically if we simply run the tests longer, and so smooth out the effects of random coincidence.

Back to top  Back to 100th Prediction Summary

Table of Day-to-Day Events in the Test

Key

• Date - The date we issued this prediction, pertaining to the next business day.
• Buy Today? - Did we recommend buying gas that day ("yes"), or waiting to buy gas the next day ("no")?
• Avg. Price - What was the second-lowest gas price in Atlanta? This is the price on the date given, not the predicted price for the next day.
• Sub. Gas in Tank - How much gas did our subscriber have in his tank at the end of that day, when he received his Gas Predictor newsletter?
• Sub. Bought - How much gas did our subscriber buy that day, whether we recommended it or not?
• Sub. Paid - How much did our subscriber pay for the gas he bought that day?
• Ctl. Gas in Tank - How much gas did our control driver have in his tank at the end of that day?
• Ctl. Bought - How much gas did our control driver buy that day?
• Ctl. Paid - How much did our control driver pay for the gas he bought that day?

Back to top  Back to 100th Prediction Summary