Welcome to the final table of the World Series of Poker. You’ve survived five grueling days of Texas Hold-Em poker with over five thousand players - and are now face to face (heads-up) against the chip leader. You look at your two cards as the dealer lays out the community cards (the flop). Your opponent looks at you, smiles, and calmly says, “I’m All-In” – What to do?
You’re only a few weeks away from your performance review and your boss calls you into her office. She asks, “If I assign you to lead a special project, can you finish it in ten months?"You have hear that this project will probably take at least one year to finish. What do you do?
Whether you’re one “call” away from millions of dollars or one “Okay” away from certain failure - the answer to both is the same – you do nothing until you calculate your probability of success and then decide if it is within your risk range.
Let’s examine the project example first.
Probability should be seen from two perspectives – worst case (conservative) or optimal case (very liberal). But let’s start in the middle first; Fifty percent (50%) will always be a mid-range estimate. This is where actual outcomes for your project are most likely to lie, and so this should generally be used for internal purposes. The project may take less time or more time, but generally this is the estimated date where the project is expected to come in. You express this by saying “I have 50% confidence in my estimate.” If greater certainty is desired (what most managers want) then more schedule will have to be allowed to improve the probability of that outcome.
Probability goals other than 50% are most often used to support business decisions. If you are bearing risk, such as a fixed price contract, then you’ll want to take precautions against a poor outcome and choose a higher probability, something near 80 or 90 percent. If you are not bearing risk, like on an R&D project, then the risk exposure is less and you can propose a bolder estimate at a lower probability setting far below 50 percent.
Determining confidence levels come in two varieties that are often confused as the same thing– Effort Probability and Schedule Probability. Effort identifies how many “hours” will it take to complete the project, Schedule is how “long” it will take. In our example, as you’re sitting across the table from the boss, you’re being asked to consider the probability of meeting a schedule! Effort – Shmeffort – can you finish this project in eight months? (We’ll save the effort discussion for another paper) Before you give an answer there are two things you should do:
1. Hold your breath and count to ten
2. Ask one very simple but strategic question – “How conservative of an answer do you want?”
You’re now likely to hear one of three answers – and with each answer you’ll know the probability required of your estimate to make her happy. If she says:
a) “Make sure your estimate is very conservative,” then what she really means is, “I need at least an 80 - 90 percent probability the schedule will be completed on time.” Ah-oh – Time to squirm in the chair – Your one year estimate is already way too low.
b) "Let's show some optimism.” Whew! She is only asking for about a 20 – 30 percent probability – Ten months might be possible after all.
c) "Give me a most likely estimate.” Whoops, now you’re in a pickle. She is asking for a 50 percent probability that the project will be completed on time - which for her is ten months - but for you is a full year.
Calculating confidence risk is easy using SEER for Software and it’s an excellent way to help management understand what a realistic outcome looks like, and can quickly provide an answer to all her possible probability ranges.
|
Estimate
|
20% Confidence
|
50% Confidence
|
90% Confidence
|
|
Effort Months
|
37.36
|
53.78
|
131.68
|
|
Schedule
|
10.48
|
12.01
|
16.35
|
Using the SEER for Software interactive “Schedule Risk” chart – we can see that - if our sample project requires a 90 percent probability - then it needs at least 16.35 months to complete. Likewise, if the project has only ten months to complete then it has less than a 20 percent probability of being finished on time.
So, as you can see, it doesn’t matter which probability range she gives you – SEER for Software will calculate an answer for you. And even more - you’ve helped the boss make a more educated decision. Smile, you can smell that raise-a-coming!
But, let's return to the poker example!
With two cards remaining to be dealt what is the probability that you can improve your hand and beat any pocket pair?
Your Cards: Queen of Diamonds, 9 of Diamonds
Your Opponent: Unknown
Community Cards: 2 of Diamonds, Jack of Diamonds, 10 of Spades, Unknown, Unknown
You can use a short cut to calculate the probability of your getting the card you need by multiplying your outs (the cards you need) by four. (With only one card remaining multiply by two)
|
Cards Remaining
|
Available
|
Probability
|
Odds of hitting the following hand.
|
|
Unseen Diamonds left
|
9
|
36 %
|
36% chance to hit a flush
|
|
Kings left*
|
3
|
12%
|
12% chance to hit the king high straight
|
|
Eights left*
|
3
|
12%
|
12% chance to hit the straight from the eight
|
|
Total Outs
|
15
|
60%
|
Confidence your hand will improve
|
(* Don’t double count the Diamond Eight or Diamond King)
What should you do? Once you perform this math in your head you will have a calculated decision to make. Are you willing to risk your remaining chips with just sixty percent confidence that your card will hit? If you miss – you lose.
I don’t know about you – but with a 60 percent probability to double up my chip stack- “I’m All In!”
David DeWitt, Senior Consultant, Galorath Incorporated. Galorath provides Cost Estimating tools for estimating software projects, manufacturing projects and estimating IT projects.
Mail this post
Technorati Tags: all in, calculation, poker, probability, software, texas hold-em
