Because circle track (and other) engines must provide power over a range of rpm, particularly at higher speeds, it becomes necessary to initiate spark in advance of TDC piston position. By correctly viewing controlled combustion as a process and not an event, you'll understand it requires a period of time. Consequently, spark timing is "advanced" to compensate for the time that's not available for combustion over a span of engine speeds. Very simply stated, you might call it a "Kentucky windage" approach allowing for a natural delay in the combustion event. However, the more timing advance that's applied, the greater the "negative" torque applied to the crankshaft before TDC on the compression stroke. It's this counter-rotational force that contributes to a net reduction in positive torque. Speed up the burn and less spark advance is required, all else being equal. So by mere acceleration of the combustion process, less work is lost trying to rotate a crankshaft in the wrong direction.
Technically speaking, we're dealing with an engine's IMEP (indicated mean effective pressure) characteristics. As derived from a P-V (pressure-volume) diagram, this condition can be determined by computing the quantity of work done on a piston per the piston's swept volume. In more general terms, it's an "average" of the work done before and after TDC piston position on the power stroke. Even more simply stated, IMEP increases as pre-TDC pressure is reduced, and that's a desirable objective. By whatever means, if the combustion process is made more rapid, less spark advance is required and IMEP can be increased by the reduction of pre-TDC force on a piston. Engines built in this fashion tend to exhibit sharper throttle response, especially under load, a partial result from having reduced negative torque on the crankshaft.
So, as an engine builder, how do we achieve these benefits? The ways are multiple, but you may want to consider some of the following. First, doing what you can to enhance fuel atomization is helpful. Even in the best of circumstances, certainly by comparison to fuel injection "spray" efficiencies, carburetors don't do a good job of creating finely atomized particles of fuel. Atomizing a solid stream of fuel by the "shearing" action of air moving through a low-pressure region is problematic at best, but good booster efficiency and downstream flow surface roughness in the intake track can help.
Making certain you've optimized flow pressure distribution patterns and extremes in pressure conditions (aided by wet-flow studies) can also provide improved fuel atomization. Remember, we're attempting to achieve and then allow the transportation of well-mixed air/fuel charges all the way into the combustion space while minimizing both fuel/air separation and increased fuel particle size. Homogeneity from carburetor throughout the "burn" is key.
You can also address the problem by creating certain "mixture motion" patterns, not unlike those referenced at the beginning of this column. Whether it's so-called swirl, tumble, or some combination thereof, it's important not to allow such processes to centrifuge fuel out of suspension with air. Deviation from linear flow conditions can also reduce net flow, so a balance between enhancing air/fuel charge conditioning and causing reductions in net volumetric efficiency is necessary.
Further, despite some opinions to the contrary, experience has shown there are benefits from "reading" combustion surfaces and skillful pressure mapping of velocity profiles by way of Pitot tube measurements. Analyzing post-combustion patterns can point to areas of high or low (rich or lean) fuel conditions that existed during the burn. Proponents of wet-flow analysis also have reasons to believe their methods are helpful. But regardless of how you may attempt to determine and tailor beneficial flow dynamics on the inlet side of an engine, in reality we're dealing with unsteady, interrupted flow that create conditions you can't precisely duplicate on a flow bench, dry or wet.
Finally, let's wrap up this discussion with some summary comments. As previously mentioned, high levels of atomization efficiency during combustion contribute to a rapid burn. The faster the combustion process (in a controlled fashion that excludes both pre-ignition and detonation), the later it can be begun. By creating an environment that allows maximum combustion pressure with less spark timing, torque applied on the piston's down-stroke is increased, accordingly. This condition can be created without any attending changes to air/fuel charge ratio, valve timing, or mechanical compression ratio ... although each of these can be "adjusted" to enhance the results of a faster burn. And in most cases, especially during engine dyno testing, you'll discover a side benefit of this approach is lower b.s.f.c. without a corresponding reduction in power. All else being equal, that can translate into quicker lap times.