A bubble is typically defined as ìtrade in high volume at prices that are considerably at variance from intrinsic valueî, which is compatible both with higher or lower prices, whereas the metaphor suggests something asymmetric that gradually inflates rather than deflates. Whether bubbles are symmetric or not is an empirical question; the historical record appears to be that they are asymmetric, but the literature is unclear on this point. In addition, in some types of market such as property, it is unclear that an ìintrinsic valueî exists in the same way that it does for e.g. stock markets. Furthermore, not all types of market are prone to bubbles. Established product markets (goods and services) do not form bubbles, whereas they are frequent in property and financial sectors, as well as with new products and also with primary products (e.g. minerals). This suggests that markets with a stable relationship with costs ñ ìcost-tetheredî markets ñ are not bubbles-prone, whereas those that lack this characteristic ñ ìfree-floatingî marketsî ñ are. In cost-tethered markets, the information required to set the going price is derived from the unit cost, whereas with free-floating markets another type of information is required. One possibility is that a recent price trend is extrapolated into the future. This creates a sellersí market in the case of a rising trend, or a buyersí market with a falling one. This situation can be modelled very simply by extending a standard market model to include additional terms that represent the effect of the trend extrapolation on the current price. It can be shown that under certain conditions, instead of the standard situation in which the quantity demanded falls with increasing price, this is reversed so that a perceived rapid upward price trend leads to increased demand. Similarly, from the sellersí viewpoint, an intensely rising trend can lead to a lower rather than the usual higher willingness to sell. These would together lead to a self-fulfilling and thus self-perpetuating tendency, whereby the trend is accentuated, a bubble equilibrium, which is inherently unstable. The maths shows that this can occur only in a rising market. The model therefore suggests that the definition of a bubble should be asymmetric. The causal processes underlying this model do not depend on radical departure from rationality, as with recent authors who have postulated ìemotionî, ìgroup-thinkî, ìoptimismî or ìpanicî. Rather, a form of situational rationality is postulated, that preserves the element of deliberate forward-oriented calculation.