The landscape of sports odds advisory services breaks down broadly into three categories:
a)Sites offering all manner of statistical data covering teams and individuals, which while useful and endlessly interesting, serve essentially as reference tools from which the user must glean their own conclusions. Information is indeed power, but without a framework for how to use it, the panoply of statistical data can be bewildering...if not misleading. And even those with an advanced knowledgebase may find it prudent to get a ‘second opinion.'
b) Simulation engines ("SIMs") offering their own brand of what could broadly be characterized as Monte Carlo scenario analysis. This is a solid method of building odds-based comparisons, particularly for higher scoring sports like basketball where the variance between sampling estimates and actual score outcomes is much smaller in percentage terms than, for instance, in hockey or baseball. As such, simulators are terrific for picking sides in basketball. They can also be useful in estimating the impact of single player substitutions, where other variables remain constant, such that the impact of shifting amongst both the existing and new players can be sampled believably. But as with almost any method, it has some limitations. In sports where single-player substitutions cause a broader shift in player assignments, or where play-calling or strategies shift due to substitutions, SIMs may struggle with game-specific matchup characteristics relating to any two team's numbers versus their schedule-to-date because there are simply too many moving parts and permutations for even thousands of simulations to estimate acceptably. Furthermore, the selection set of simulation variables, including (if at all) any filtering of what are thought to be redundant or innocuous data points which do not translate into team performance, can be key to figuring lines and odds. This includes weighing the degree to which certain statistics are ‘baked-in' to others, or are more relevant as standout ‘fantasy' measures than to line dynamics. SIM purists may assert that the very value of simulation exists in the idea that one cannot [or should not attempt to] filter the data set, but the very fact that human performance is NOT random, makesLineAdvisor wary of that concept. There are a number of SIM's offerings online, and the proof of our above ‘selection set' assertion is readily apparent in the sizable difference among these services, as far as the quality of their methods and performance goes.
c)A large number of ‘qualitative' picking services whose abilities range from good, to poor, to unethical. And even when good, applying qualitative judgments can lead to a standard deviation of performance which is wildly inconsistent, as they require the ‘picker' to be particularly attuned to the source of their own decision points, and what impacts them. As a result the subscriber is in the difficult position of having to spend as much time assessing the quality of his advice, as he could otherwise have spent doing his own analysis.