Consider a multi-period model with $t=0,...,T$. Suppose there is a bond with $B_0=1$ and $B_t=(1+R)^t$ and a stock with $S_0=s_0$ and
$$ S_{t+1}=S_t\,\xi_{t+1}, $$
with $\xi_t$ iid random variables. I indicate with $(\alpha_t,\beta_t)$ the predicable components of the portfolio. I have found two different definitions of self-financing portfolio that I would like to re-conciliate. In the book by Tomas Bjork ("Arbitrage Theory in Continuous Time") it is said that the value of the portfolio at time $t$ is
$$ V_t^{(\alpha,\beta)} = \alpha_t\,S_t+\beta_t\,(1+R) $$
and the self-financing condition is expressed as
$$ \alpha_t\,S_t+\beta_t\,(1+R) = \alpha_{t+1}\,S_t+\beta_{t+1}. $$ This is quite intuitive form me since, in Bjork, $\alpha_t$ (resp. $\beta_t$) is, by definition, the amount of money we invest in the stock (resp. in the bond) at time $t-1$ and keep up to time $t$. So if I buy $\beta_t$ unit of the bond at time $t-1$ then I gain $\beta_t\,(1+R)$ at time $t$.
Nevertheless, in the book by Andrea Pascucci ("PDE and Martingale Methods in Option Pricing") it is said that the value of the portfolio is
$$ V_t^{(\alpha,\beta)} = \alpha_t\,S_t+\beta_t\,B_t = \alpha_t\,S_t+\beta_t\,(1+R)^t $$
and the self-financing condition is expressed as
$$ V_t^{(\alpha,\beta)} =\alpha_t\,S_t+\beta_t\,(1+R)^t = \alpha_{t+1}\,S_t+\beta_{t+1}\,(1+R)^t. $$ Pascucci define $\alpha_t$ (resp. $\beta_t$) as the amount of the asset $S$ (resp. of the bond $B$) held in the portfolio during the period $[t-1,t]$. Are the two definitions equivalent? I am pretty sure that the the solution is in the fact that in Bjork it is defined as the amount invested while in Pascucci as the amount held. Nevertheless I miss which kind of relationship is in between the two.